<?xml version="1.0" encoding="UTF-8"?><rss xmlns:dc="http://purl.org/dc/elements/1.1/" xmlns:content="http://purl.org/rss/1.0/modules/content/" xmlns:atom="http://www.w3.org/2005/Atom" version="2.0" xmlns:cc="http://cyber.law.harvard.edu/rss/creativeCommonsRssModule.html">
    <channel>
        <title><![CDATA[Stories by Felix M. Simon on Medium]]></title>
        <description><![CDATA[Stories by Felix M. Simon on Medium]]></description>
        <link>https://medium.com/@felixsimon?source=rss-41832dfb135a------2</link>
        <image>
            <url>https://cdn-images-1.medium.com/fit/c/150/150/1*_wNzidEACwBqxrGyyLl77Q.jpeg</url>
            <title>Stories by Felix M. Simon on Medium</title>
            <link>https://medium.com/@felixsimon?source=rss-41832dfb135a------2</link>
        </image>
        <generator>Medium</generator>
        <lastBuildDate>Mon, 16 Nov 2026 01:24:51 GMT</lastBuildDate>
        <atom:link href="https://medium.com/@felixsimon/feed" rel="self" type="application/rss+xml"/>
        <webMaster><![CDATA[yourfriends@medium.com]]></webMaster>
        <atom:link href="http://medium.superfeedr.com/" rel="hub"/>
        <item>
            <title><![CDATA[Thoughts on the Value of News and Sustainable Media Models in a World of AI]]></title>
            <link>https://felixsimon.medium.com/thoughts-on-the-value-of-news-and-sustainable-media-models-in-a-world-of-ai-7e8ab6e186b3?source=rss-41832dfb135a------2</link>
            <guid isPermaLink="false">https://medium.com/p/7e8ab6e186b3</guid>
            <category><![CDATA[ai]]></category>
            <category><![CDATA[journalism]]></category>
            <category><![CDATA[news]]></category>
            <category><![CDATA[democracy]]></category>
            <dc:creator><![CDATA[Felix M. Simon]]></dc:creator>
            <pubDate>Sat, 20 Dec 2025 10:56:26 GMT</pubDate>
            <atom:updated>2025-12-20T10:56:26.093Z</atom:updated>
            <content:encoded><![CDATA[<figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*OvvIE_7qVuxK5zwdOKELrw.jpeg" /></figure><p><em>The below are slightly copy-edited remarks I delivered to the Ditchley Summit on </em><a href="https://theditchleyfoundation.my.salesforce.com/sfc/p/#8d000001zOYI/a/Q5000007EmC1/._QVRMQHaL.DpCscJph.1rvrVqWSkNlzcQnTJPJAaLs"><em>The Role of the New Fourth Estate in Renewing Democracy </em></a><em>in early December.</em></p><p>Thank you. I’ve been asked to say a few words on this question: ‘<em>What might sustainable media models for rigorous news gathering and reporting be in a world where AI may control the attention economy — and what are the attributes of news that AI cannot adequately replace?’</em></p><p>Usually, I find it’s good to start with analysing the premise and assumptions behind such a question before we try and answer it.</p><p>Here these are as follows:</p><ol><li>We assume that news and the function of news gathering and reporting are a beneficial good for the public and by extension democratic life.</li><li>We assume, second, that AI inserts itself somewhere in this equation. The question then is where, in what way, and with what effect?</li><li>We assume that there are attributes of news that AI (or its use by other actors) cannot adequately replace. If so, which are these?</li></ol><p>So let me take these apart before I deliver my answer. First, is newsgathering and reporting a beneficial good and if so why?</p><p>Especially in Western political thought, news and information is by default framed as a normative good because it is seen as providing the conditions for one of the core tenets of democracy, namely self-rule (so the notion that people should choose who governs them), which is seen as a superior form of decision-making in complex social systems. Self-rule here is predicated on the idea that people can make better decisions based on good information from a plurality of sources. This is not to say that people always do the same or that decisions cannot be made without good information — but it is something that no small number of people can agree on.</p><p>There are lots of other functions news has: connecting elites with the public in a conveyor belt fashion, so various parts of the social system can know about each other; providing a sense of cohesion and connection. However, for the sake of our argument, let’s focus on a minimalist definition of news’ role in democracy provided by Rasmus Nielsen: news has the function of providing people with relatively accurate, accessible, diverse, relevant, and timely independently produced information about public affairs.</p><p>We have relatively good empirical evidence that news does these things at least some of the time, if not always in a perfect way. It helps people be more informed (surprise), and it can also increase resilience to misinformation. Higher news consumption is also correlated with better political knowledge, higher political participation, higher societal trust in each other, and potentially lower levels of polarisation.</p><p>So, in a nutshell, we have some good normative and empirical arguments that news gathering, reporting, and the distribution of the same through news from a plurality of sources is a good thing.</p><p>Which brings us to the second point: the role of AI.</p><p>Even before AI, news across countries has faced strong competition for attention and money for the better part of the last two decades. This has multiple causes which are worth pointing out just briefly.</p><p>We have seen the rise of a high-choice media environment through digital media, giving people ability to be pickier and leading to more competition for attention.</p><p>We have seen a disruption of old business models of news, which were mostly advertising funded.</p><p>We have also seen an unwillingness or inability to innovate and adapt to changing modes of reception and consumption on parts of the news.</p><p>We also seen a growing share of people having declining trust in news, going along with rising disinterest in news and an unwillingness to pay for it. This trend is stronger among younger audiences who find social media platforms more accessible and engaging, often gravitating towards formats that prioritise personalities over traditional news sources.</p><p>Finally we shouldn’t forget other factors, some of which have led to some of these things: political attacks against the press, forms of lawfare, or funding cuts to public service news.</p><p>With AI, we have seen now further upheaval. As plenty of empirical evidence, including some of my own work and research at the Reuters Institute has shown, technology companies have already become the main gateways to news and information for audiences around the world. So they have a strong control over distribution and access to news. For news organisations that did work to a degree because they were sent traffic which allowed them to do advertising but also turn visitors into subscribers. Granted, this didn’t work for all, and was difficult at the local level, but it did provide some sort of uneasy equilibrium.</p><p>With AI, technology platforms are now moving more strongly into the production of information itself with some of the products built on the back of this — and trained and integrating journalism along the way — increasingly acting as complements or even substitutes for what news used to provide.</p><p>We find in our research that a majority across countries now, for example, see AI-generated summaries in search. But only about a third say they always or often click the links in AI search responses. Even though the overall picture is difficult to ascertain, other studies from, e.g. Pew in the US, come to similar conclusions, and web tracking data finds declining traffic to publishers.</p><p>So taken together, newsgathering and reporting as a business have been in a difficult position for a variety of reasons for a long time, but the use of AI opens up new opportunities, but also new competition for news, because it increases control in one sector over the structuring of information, the conditions for attention and the ability to make money on the back of this.</p><p>Which gets us to the third point: We assumed in the question that there are attributes of news that AI cannot adequately replace. And we also assume that news is broadly a good thing for society.</p><p>So if we accept the last premise, what exactly are the attributes — if any — that AI cannot replace in terms of what news does? And to finally give an answer, what might sustainable media models look like given the broader situation I outlined?</p><p>I submit that AI on its own — so where it is used as something within products of platforms or as standalone systems — is currently ill-suited to provide some of the functions set out earlier when I defined the minimal role of news following Nielsen as providing us with accurate, accessible, diverse, relevant, and timely independently produced information about public affairs.</p><p>To be clear, it can do some of this! In terms of providing accurate information, systems still struggle, although they have improved; however, this area also isn’t as well studied at scale as we’d wish it to be. In terms of accessible information, if we define this as providing news and information in ways that are suitable to people with different needs, again, this is something these systems can do (and are used for this purpose as our research shows), although this heavily depends on language and context. Where it gets thornier is diversity and relevance. On the diversity element, there is some early data to indicate that they are skewed towards news and information from national sources but struggle locally and across the Global South. But in terms of relevance, data that a growing number of people use it for news and the main use case now is for information retrieval suggest something is up here, too.</p><p>Where AI — and its use by non-news actors — struggles most is with the ‘independently produced information part’. Because while the act of generating news-like information at inference time might be equated with production, AI systems as of now do not produce the underlying information they use for it themselves. These systems also cannot yet report, go out, or solicit sources in the field. They do not decide to attend a council meeting, knock on doors, cultivate sources, or file freedom of information requests. Instead, they have to rely on a handful of licensing agreements with news organisations in some cases and crawling and scraping of information, including news, in the case of others to do this work. This might change in the future as we move into stronger forms of agentic and embodied AI or even AGI, with more sensors and all the like potentially enabling similar work. I don’t want to rule out this possibility but I am an empiricist at heart and for now at least, this is still a bottleneck.</p><p>They also do not take responsibility in the way human newsrooms do. Editors and reporters can be named, questioned, praised, or blamed. They operate under professional norms and ethical codes. An AI system may apologise when it gets something wrong, but it cannot be held accountable in the same way.</p><p>Another attribute (again for now) is editorial judgement. News is not just any information; it is a set of decisions about what matters, for whom, and why now, ideally grounded in local knowledge and in relationships with communities. A model, by contrast, is an optimisation engine for objectives chosen by its owners — objectives that may prioritise engagement or revenue over democratic value (although one can rightly argue that some news organisations also have a sketchy record on the latter aspect).</p><p>I would also submit that most technology companies, and most other actors deploying AI, are not in the business of independent scrutiny of power. This is not because it could not be done by them. One could flippantly argue that Amazon, Google, Microsoft, or OpenAI could easily start this if they wanted to, and it wouldn’t even cost them much. The yearly operating costs, for example, of the New York Times were about 580 million US dollars in 2024. The revenue of Alphabet was 350 billion US dollars in the same year, Microsoft’s was at about $245 billion. Google spent roughly $53 billion in 2024 on AI infrastructure alone, and for Microsoft, it was roughly $88 billion. In other words: they could do something that comes close to news, and it would be a rounding error in their capex. But this is not what technology companies see themselves as and not their raison d’être. And fair enough! But it also goes to show that newsgathering and reporting is not something that will magically appear out of nothing with AI.</p><p>So if we want sustained, critical reporting because we think it delivers some public value — but also because it is in parts needed to improve the delivery of news and real-time information regarding public affairs via AI — it is safest to assume that it will have to come from actors whose primary mission is journalism (or something like it), not from those whose primary mission is advertising, cloud computing, or building general purpose technologies.</p><p>So what to do then and by whom? None of the following suggestions rule each other out, instead they would best work in tandem:</p><p>First, until or unless technology companies get into the news reporting game (which is unlikely), better revenue-sharing models between news and technology companies that accept and value that news makes a contribution to information provision within these systems and where people use them would be needed. This also includes greater cooperation on adequately representing news within those systems if we work from the basis that it is a good worth preserving. There is willingness to pay for other things and spend big; it’s not impossible to add a bit more and pay for training data used at inference time. Some of this is already happening. And while this is not going to save the news — and while we need more independent work to correctly assess what the right value of such deals should be — it presents a start.</p><p>Second, there is a role for public policy. If we accept that independent news is a public good under pressure, then there are ways to fund this that are publicly acceptable, prevent media capture and can be tied to specific conditions so that they do not amount simply to handouts or strengthening incumbents. This is particularly relevant at the local level, where markets often fail. My colleague Rasmus Nielsen recently outlined how this might be done for Denmark but I think his lessons apply more broadly.</p><p>Third, news organisations themselves will have to lean into what they can do that AI cannot and deliver news in ways that speak to people and their needs. Despite proclamations to the contrary, we still see news organisations not delivering on this and producing news that adds little value to anyone’s life. Sustainable models will probably combine multiple revenue streams — subscriptions, memberships, philanthropy, and commercial income — with smarter use of AI to deliver on the distinctively human parts of journalism that people have been shown to value.</p><p>***</p><p><em>Because this was a speech, hashed out over night I refrained from linking to some of the sources for lack of time, but for further reading here are some of the pieces I drew from:</em></p><p>Altay, S., Nielsen, R. K., &amp; Fletcher, R. (2024). News Can Help! The Impact of News Media and Digital Platforms on Awareness of and Belief in Misinformation. <em>The International Journal of Press/Politics</em>, <em>29</em>(2), 459–484. <a href="https://doi.org/10.1177/19401612221148981">https://doi.org/10.1177/19401612221148981</a></p><p>Humprecht, E., Esser, F., &amp; Van Aelst, P. (2020). Resilience to Online Disinformation: A Framework for Cross-National Comparative Research. The International Journal of Press/Politics, 25(3), 493–516. <a href="https://doi.org/10.1177/1940161219900126">https://doi.org/10.1177/1940161219900126</a></p><p>Jungherr, A., Rivero, G., &amp; Gayo-Avello, D. (2020). <em>Retooling Politics: How Digital Media Are Shaping Democracy</em>. Cambridge University Press. <a href="https://doi.org/10.1017/9781108297820">https://doi.org/10.1017/9781108297820</a></p><p>Nielsen, R. K. (2017). The One Thing Journalism Just Might do for Democracy. <em>Journalism Studies</em>, <em>18</em>(10), 1251–1262. <a href="https://doi.org/10.1080/1461670X.2017.1338152">https://doi.org/10.1080/1461670X.2017.1338152</a></p><p>Simon, F. M. (2022). Uneasy Bedfellows: AI in the News, Platform Companies and the Issue of Journalistic Autonomy. <em>Digital Journalism</em>, <em>10</em>(10), 1823–1854. <a href="https://doi.org/10.1080/21670811.2022.2063150">https://doi.org/10.1080/21670811.2022.2063150</a></p><p>Simon, F. M. (2025). Rationalisation of the news: How AI reshapes and retools the gatekeeping processes of news organisations in the United Kingdom, United States and Germany. <em>New Media &amp; Society</em>, 14614448251336423. <a href="https://doi.org/10.1177/14614448251336423">https://doi.org/10.1177/14614448251336423</a></p><p>Simon, F. M., Nielsen, R. K., &amp; Fletcher, R. (2025). <em>Generative AI and News Report 2025: How people think about AI’s role in journalism and society</em>. Reuters Institute for the Study of Journalism. <a href="https://doi.org/10.60625/RISJ-5BJV-YT69">https://doi.org/10.60625/RISJ-5BJV-YT69</a></p><p>Nielsen, R. K. (2025, November 12). <em>From taxes to news: How Denmark is rethinking public funding for private publishers</em>. Reuters Institute for the Study of Journalism. <a href="https://reutersinstitute.politics.ox.ac.uk/news/taxes-news-how-denmark-rethinking-public-funding-private-publishers">https://reutersinstitute.politics.ox.ac.uk/news/taxes-news-how-denmark-rethinking-public-funding-private-publishers</a></p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=7e8ab6e186b3" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[My Favourite Books of 2025]]></title>
            <link>https://felixsimon.medium.com/my-favourite-books-of-2025-611953ee9a8e?source=rss-41832dfb135a------2</link>
            <guid isPermaLink="false">https://medium.com/p/611953ee9a8e</guid>
            <category><![CDATA[christmas]]></category>
            <category><![CDATA[books]]></category>
            <category><![CDATA[ai]]></category>
            <category><![CDATA[reading]]></category>
            <dc:creator><![CDATA[Felix M. Simon]]></dc:creator>
            <pubDate>Tue, 16 Dec 2025 13:31:35 GMT</pubDate>
            <atom:updated>2025-12-16T21:18:40.353Z</atom:updated>
            <content:encoded><![CDATA[<figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*5EOggRUkWGLzS3ok7akxew.png" /><figcaption>Just as last year, another Oxford-variation on Pieter Breughel’s “Hunters in the Snow”, produced by Google Gemini’s Nano Banana.</figcaption></figure><p>Every year, I list a couple of books I particularly enjoyed over the last year (see also the <a href="https://felixsimon.medium.com/o-come-all-ye-curious-6ff9c3d3c44b">2019</a>, <a href="https://felixsimon.medium.com/my-favourite-books-of-2020-d3f04f1c4665">2020</a>, <a href="https://medium.com/@felixsimon/my-favourite-books-of-2021-ef13ed06eb29">2021</a>, <a href="https://felixsimon.medium.com/my-favourite-books-of-2022-c6f24f588039">2022</a>, <a href="https://felixsimon.medium.com/my-favourite-books-of-2023-1625a5ce039a">2023</a> and <a href="https://medium.com/@felixsimon/my-favourite-books-of-2024-cd6bec59b96c">2024</a> editions) and some new Christmas albums. The latter I have decided to skip for this edition (because, frankly, I couldn’t find anything new that I liked). But if you are in need of some inspiration for last minute presents, this will hopefully prove helpful (and for albums you can always check the editions from previous years).</p><p>Instead of music, I will use this space to plug one piece that I am particularly proud of: “Don’t panic (yet): Assessing the evidence and discourse around generative AI and elections” co-written with Sacha Altay came out with the <a href="https://knightcolumbia.org/content/dont-panic-yet-assessing-the-evidence-and-discourse-around-generative-ai-and-elections"><em>Knight First Amendment Institute</em></a><em> </em>earlier this year and makes a sweeping evidence based case why some of the arguments we see around AI’s impact on elections and democracy are problematic<em>.</em></p><p>Another addition this year, two recommendations for Christmas Donations: I have given to Wikipedia (<a href="https://donate.wikimedia.org/w/index.php?title=Special:LandingPage&amp;country=GB&amp;uselang=en&amp;wmf_medium=spontaneous&amp;wmf_source=fr-redir&amp;wmf_campaign=spontaneous">donate here</a>) and the WebArchive (<a href="https://archive.org/donate/">donate here</a>) because I believe in their mission and think they are worth defending.</p><p>A very happy Christmas to you all from Oxford and happy reading🎄</p><h3>Books</h3><p><strong>On Truth in Politics<br></strong><em>Michael Patrick Lynch, 2025, Princeton University Press</em></p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*FG3aXXsrnpe8r9G7iQk8mQ.jpeg" /></figure><p>Michael Patrick Lynch’s <em>On Truth in Politics: Why Democracy Demands It</em> is the kind of book that made me stop every couple of pages to scribble things in the margin. Like “basic definition of truth and why it matters” — partly because Lynch insists on starting there (which is good, I love a good “here’s what I mean by this”), and partly because it’s weirdly easy to forget how much political arguing is actually arguing about whether we even owe each other that baseline. The setup for the book is bleak but (unfortunately) familiar: democracies are wobbling, and at the same time it’s become normal to treat truth as either optional or partisan, “politics is only about power,” full stop. Lynch’s wager is that this is exactly backwards: truth isn’t some nerdy add-on to democracy; it’s one of the things democratic politics quietly requires to function at all.</p><p>One idea I keep circling in my own work is what some call “knowledge infrastructure” or “epistemic infrastructure”: the institutions, norms, and habits that make it possible for citizens to figure things out together (and not just vibe in parallel) and this also something Lynch deals with. I always find this framing helpful because it shifts the conversation away from individual “fact checking” heroics (which do exist and are needed) and toward the boring, structural stuff we only notice once it’s broken — education, journalism, science. Lynch also throws in a moral element: the idea that democratic politics depends on treating other people as <em>epistemic agents</em> — as reason-givers who deserve a basic level of respect even when they’re infuriating. If we fail to treat our opponents as fellow reasoners capable of changing their minds, rather than as enemies to be manipulated, we aren’t engaging in politics at all, but merely coercion. In other words, democracy isn’t just votes; it’s a “space of reasons” and we are all engaged in that collective enterpriese and truth is a core element of it, in some way or another. And yes, this is all very pragmatist in spirit (Dewey hovering in the background).</p><p>A key point Lynch makes (as I read it) is that there has been temptation in recent years to let politics collapse into performance, conformity, and status games — until “truth” gets demoted beneath vibes, dunks, and tribal belonging. And while this has always happened and might not be something that can ever fully be prevented, he makes a clear case for why we should be striving for the opposite. Another point he makes is that fights over facts are fights over political meaning and interpretation. That’s Lynch at his most useful, because it explains why perfectly good evidence or facts can bounce off a conversation without landing: people aren’t only contesting propositions; they’re defending identities, communities, and moral narratives (which is something I really would like the “if we only had more facts” crowd to understand: the social associations we create (the context in which we interpret facts) are just as important as the facts themselves).</p><p>What I liked most is that Lynch doesn’t try to “solve” politics by pretending it could become physics. He’s comfortable with pluralism and messiness, while still insisting that political judgments can be true or false in ways that matter…otherwise we’re just doing aesthetics with higher stakes. I finished the thinking something like: a democracy can probably limp along without truth… but it will limp, and it will get mean. If you’re tired of overheated “post-truth” clichés but still want a serious account of why reality-based disagreement is worth defending (and what has to exist <em>around</em> us for that to be possible), this is a great read.</p><p>(<em>Yes, I do realise the above is more a stream of consciousness vomiting of thoughts and not a summary. Then again, I am writing this for fun and for myself so hey ho).</em></p><p><strong>Shame &amp; A Man’s Place<br></strong><em>Annie Ernaux, 2023, Fitzcarraldo Editions</em></p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*9tZnmYU06NHS1cOM2wfLog.jpeg" /></figure><p>Have I entered my performative male era with the recommendation of Fitzcarraldo edition books?? Maybe, but to be honest I don’t care. I picked these up on my partner’s recommendation and turns out they are not just great books, but really good descriptive sociology effectively masking as autobiographical non-fiction novels. Ernaux manages to express whole inner universes with just a few sparse lines. I am not usually one turned off by literary maximalism (Henry James anyone?), but increasingly feel more drawn to a more economical approach to writing (a sign of age and ultimate death approaching and thus being secure in the knowledge that my time is limited? Too deeply annoyed by verbose academic texts that say very little with a lot? God knows). Ernaux is a master of the form — and while we will never be able to say how “true” she is to the actual course of events, this doesn’t matter all that much because she always portrays this as at best a reflection of her experience and is honest with the reader that it’s her view above all. And I dare anyone to be as unflinchingly honest, if not brutal, with their own life and that of their friends and family in their writing as she is in these. If I wrote stuff like this, my parents would probably disinherit me.</p><p><strong>AI, Automation, and War<br></strong><em>Anthony King, 2025, Princeton University Press</em></p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*-3TIb_pDWa8aSIUcjGLNHQ.jpeg" /></figure><p>In the popular imagination, the intersection of AI and warfare summons images of Terminator-style killer robots and autonomous machines making life-or-death decisions without human intervention. In his sharp and necessary and deeply researched (some 100+ interviews) new book, Anthony King, Professor of War Studies at the University of Exeter, dispels some of these sci-fi fantasies to reveal a reality that is both more mundane and more interesting.</p><p>King’s central argument is that we are witnessing the birth of a “military-tech complex,” a distinct evolution from the 20th-century military-industrial complex. The difference here is structural: whereas the old model was defined by state ownership of hardware (where governments bought tanks or jets to operate themselves) the new complex is defined by <em>access</em> to software and data services. In this ecosystem, militaries do not own the digital infrastructure they rely on; they merely lease it.</p><p>This shift means that private technology giants are no longer just distant contractors; they are increasingly indispensable participants in the kill chain. As King notes, this reliance on commercial access has led to a quasi-privatisation of national defence strategy, where tech oligarchs now also possess the power to veto military operations — exemplified by Musk’s unilateral decision to restrict Starlink access near Crimea to prevent a Ukrainian drone strike on the Russian fleet.</p><p>Contrary to the “move fast and break things” ethos of Silicon Valley, King argues that the integration of AI into warfare has produced a paradox. While AI enables data processing and targeting at the “speed of light,” it has not resulted in the swift, decisive victories predicted by early theorists. Instead, the “transparent battlefield” — where cheap drones and AI-driven sensors make movement almost impossible to hide — has forced armies to disperse and dig in. The result, seen starkly in Bakhmut, is a high-tech version of First World War trench warfare: a grinding war of attrition where to move is to be seen, and to be seen is to die.</p><p>Rather than full autonomy, the future of war in King’s telling lies in human-machine teaming, where AI handles the data-crunching drudgery while humans retain strategic judgment. However, this partnership is fraught with friction. <em>AI, Automation, and War</em> is a sober, empirical correction that takes military application of AI seriously (and argues for the significance of it) but does not succumb to some of the breathless hype surrounding it. It reminds us that while the tools of war are changing at blinding speed, the nature of war — a chaotic, political, and deeply human contest of wills — remains stubbornly the same.</p><p><strong>How Progress Ends<br></strong><em>Carl Benedikt Frey, 2025, Princeton University Press</em></p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*skRlnfpG0hVyo2a0ZZznoA.jpeg" /></figure><p>Only a couple of points on this one: First, Carl sat on my PhD committee. Second, (and hence the first point), while I have not actually read the book I will make an exception in this case because I’ve had plenty of conversations with Carl about this work and I’ve gone to two talks where he presented it, one of which I will link to (<a href="https://www.youtube.com/live/gqliq0mYzSs?si=uJskpykJTZZ90Gvv">find it here</a>). Some key takeaways:</p><p>Carl argues from history that innovation requires different approaches at different stages, with centralised systems better at scaling existing tech (think Prussia and postwar Japan successfully using state-driven approaches to adopt and scale technologies developed elsewhere) and decentralised systems better for pioneering new technological frontiers (think Britain’s Industrial Revolution and America’s Silicon Valley which initially thrived on decentralisation and openness). He also says that neither path guarantees sustained innovation. Free markets can calcify into vested interests that resist disruption (Britain), while dominant bureaucracies can lack the agility to push beyond technological frontiers.</p><p>One major takeway or warning is that the shift toward industrial policy and protectionism in the West is not without risks. In Carl’s view, without balancing centralised coordination with decentralised discovery and maintaining openness, competition, and liberalisation, current policies risk producing stagnation rather than progress — or at least that’s what I gathered from his talk. Now I will actually have to read the 600 or so pages he produced!</p><p><strong>These Strange New Minds<br></strong><em>Christopher Summerfield, 2025, Viking</em></p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*vZ_kmp-_3sW5uv6_FEHlKg.jpeg" /></figure><p>Finally, a book on AI which I have been recommending in private far and wide from my esteemed Oxford colleague Chris Summerfield. <em>These Strange New Minds </em>has made me quite a bit more bullish on AI perhaps because it makes good, factually supported arguments (something that <a href="https://www.science.org/doi/full/10.1126/science.aea3884">Chris’ work has found to be the secret of AI’s persuasive power</a>) and because it doesn’t fall into the hype camp. It also comes from someone who has spent a lot of time conducting some cutting-edge research on these systems, both within a tech company (Google), academia (Oxford) and now government (the UK’s AI Security Institute, AISI).</p><p>I recommend the book particularly to those would like to get an accessible and thorough overview of how we got to where we are with foundation models and large language models, some insights into where this all might go, and pretty sensible thoughts on some of the effects — all written in a style that often channels Douglas Adams (Chris is pretty witty in person and thankfully his editors seem to have realised that this is worth preserving in this book). Especially one of Chris’s core arguments, that we are dealing with something in the very latest models that is different and bizarre and maybe even something mind-like is something I find increasingly hard to resist.</p><p><em>Dr</em> <a href="https://www.felixsimon.net/"><em>Felix Simon</em></a><em> is a Research Fellow in AI and News at the Reuters Institute for the Study of Journalism, a Research Associate at the Oxford Internet Institute, and a Junior Research Fellow in Politics at Corpus Christi College, all at the University of Oxford.</em></p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=611953ee9a8e" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[BII Symposium 2025 — Reckoning with AI Agents]]></title>
            <link>https://felixsimon.medium.com/bii-symposium-2025-reckoning-with-ai-agents-0d38070bbafd?source=rss-41832dfb135a------2</link>
            <guid isPermaLink="false">https://medium.com/p/0d38070bbafd</guid>
            <dc:creator><![CDATA[Felix M. Simon]]></dc:creator>
            <pubDate>Wed, 14 May 2025 13:35:17 GMT</pubDate>
            <atom:updated>2025-05-23T08:59:01.142Z</atom:updated>
            <content:encoded><![CDATA[<h3>BII Symposium 2025 — Reckoning with AI Agents</h3><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*5HwfxoaZtjSThaswgoTX6Q.jpeg" /></figure><p>A growing number of researchers have identified AI agents — AI systems with the capacity to execute complex, multistep workflows without human intervention — as a significant innovation. Various technology companies are developing such agents, and some early versions have already been launched, including OpenAI’s Operator, an AI agent capable of autonomously performing tasks like online purchases and social media updates. Similarly, Anthropic’s Claude model can (in theory) control users’ computers to automate repetitive tasks.</p><p>The emergence of more cost-effective AI models, which offer greater capabilities at a fraction of previous costs, is leading to a flurry of research labs and companies developing similar systems — with the aim of integrating them into our private and public lives, into the workplace, and other infrastructures, including public services. These developments are still at an early stage, but the pace is increasing. We have a responsibility to proactively understand and address the societal and ethical implications of these systems.</p><p>To investigate recent developments and to gauge the opinions of experts and academics, we are hosting a research symposium at Balliol College in Trinity Term 2025 (Saturday, 24th May 2025).</p><p>Bringing together a diverse group of speakers, researchers, and students, the symposium will place the latest developments around AI agents in a broader societal context and evaluate their implications for our social and academic lives, the world of work, and democracy.</p><p>The event is a collaboration between <a href="https://www.oxford-aiethics.ox.ac.uk/linda-eggert">Dr Linda Eggert</a> (Faculty of Philosophy &amp; Institute for Ethics in AI) and <a href="https://www.felixsimon.net/">Dr Felix M. Simon</a> (Reuters Institute for the Study of Journalism and Oxford Internet Institute) and is generously supported by the Balliol Interdisciplinary Institute.</p><h3>Event Information</h3><p>The event takes place on <strong>Saturday, 24th May 2025 at Balliol College</strong>, starting at 12.15pm and ending at 5pm.</p><p>Spaces are very limited and we cannot guarantee that you will receive a spot. To <strong>register your interest</strong>, please use the following link: <a href="https://forms.office.com/e/PpA86ZW1LQ">https://forms.office.com/e/PpA86ZW1LQ</a>. We will notify those successful by the end of day on Wednesday, 21st May 2025.</p><p>Please only sign-up if you are confident that you can attend. (We reserve the right not to invite unexplained “no shows” to future events.)</p><p>The symposium takes place under the Chatham House Rule. We will publish a short report, summarising the findings, in the weeks following the event.</p><h3>Timeline</h3><h4>Welcome Remarks (12.30–1pm)</h4><h4>What are AI Agents? (1pm-2pm)</h4><p>The <strong>first panel will focus on AI agents </strong>in general and set the scene for the rest of the day. We will explore how AI agents work, how they are currently used and how they might be used in the future, who develops them and for what purposes, how far along we are on the transition from current AI systems to more autonomous agents.</p><p>Speakers:</p><ul><li><strong>Prof Rahul Santhanam, </strong>Professor of Computer Science at Oxford University and a Tutorial Fellow at Magdalen College</li><li><strong>Jonathan Rystrøm, </strong>DPhil student in Social Data Science at the OII. His research focuses on socio-technical evaluations of multimodal human-AI interactions</li></ul><h4>First Coffee Break (2pm-2.30pm)</h4><h4>AI Agents and the World of Work (2.30pm-3.30pm)</h4><p>The <strong>second panel will focus on AI agents’ impact on work</strong>: where and how they might aid and augment work in various industries and sectors, including in academia, banking, and medicine. The panel will also address potential resulting risks to human livelihoods, autonomy, and dignity.</p><p>Speakers:</p><ul><li><strong>Gill Whitehead</strong>, Executive Director at NatWest Group plc, Informa plc, the British Olympic Association, and Chair of the next Women’s Rugby World Cup (England 2025). Visiting Policy Fellow, Oxford Internet Institute. Former Group Director, Online Safety, at Ofcom</li><li><strong>Dr Adam Mahdi, </strong>Group Lead, Reasoning with Machines AI Lab (RML) and Departmental Lecturer, Oxford Internet Institute. Fellow of Wolfson College, University of Oxford</li></ul><h4>Second Coffee Break (3.30pm-4pm)</h4><h4>AI Agents, the Information Ecosystem, and the Future of Democracy (4pm-5pm)</h4><p>The <strong>third, and final panel, will consider the role of AI agents in information ecosystems and their impact on democracy</strong>: How might AI agents affect information consumption, including news, and general informedness? How real and serious is the risk of people being misinformed? In what ways might they be used to promote democratic values, such as free speech, inclusion, equality, pluralism, and diversity?</p><p>Speakers:</p><ul><li><strong>Professor Scott Hale</strong>, Associate Professor and Senior Research Fellow at the Oxford Internet Institute and a Fellow of the Alan Turing Institute</li><li><strong>Chris Moran</strong>, Head of Editorial Innovation and the Editorial lead on generative AI at <em>The Guardian</em></li><li><strong>Professor Zeynep Pamuk</strong>, Associate Professor of Contemporary Political Theory and Professorial Fellow at Nuffield College, University of Oxford</li></ul><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=0d38070bbafd" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[My Favourite Books of 2024]]></title>
            <link>https://felixsimon.medium.com/my-favourite-books-of-2024-cd6bec59b96c?source=rss-41832dfb135a------2</link>
            <guid isPermaLink="false">https://medium.com/p/cd6bec59b96c</guid>
            <category><![CDATA[germany]]></category>
            <category><![CDATA[free-will]]></category>
            <category><![CDATA[technology]]></category>
            <category><![CDATA[ai]]></category>
            <category><![CDATA[books]]></category>
            <dc:creator><![CDATA[Felix M. Simon]]></dc:creator>
            <pubDate>Fri, 13 Dec 2024 19:16:42 GMT</pubDate>
            <atom:updated>2024-12-13T19:16:42.969Z</atom:updated>
            <content:encoded><![CDATA[<figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*WtvcUHsx0mZIv88dQAejoQ.jpeg" /></figure><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*XbLtbdUZvn_-aTvRPwlnOg.jpeg" /><figcaption>A Google Gemini Oxfordshire-variation on “The Hunters in the Snow” by Pieter Breugel the Elder</figcaption></figure><p>Every year, I list some of my favourite new Christmas albums as well as a couple of books I particularly enjoyed over the last year (see also the <a href="https://felixsimon.medium.com/o-come-all-ye-curious-6ff9c3d3c44b">2019</a>, <a href="https://felixsimon.medium.com/my-favourite-books-of-2020-d3f04f1c4665">2020</a>, <a href="https://medium.com/@felixsimon/my-favourite-books-of-2021-ef13ed06eb29">2021</a>, <a href="https://felixsimon.medium.com/my-favourite-books-of-2022-c6f24f588039">2022</a>, <a href="https://felixsimon.medium.com/my-favourite-books-of-2023-1625a5ce039a">2023</a> editions). Admittedly, I was fairly close to skipping it this year but had decided otherwise after three people in a row asked me about this year’s edition. So here we go again, I guess. A very happy Christmas to you all from Oxford and happy reading and listening 🎄</p><h3>Books</h3><p><strong>Caledonian Road<br></strong><em>Andrew O’Hagan, 2024, Faber &amp; Faber</em></p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*Ay7EZuSf_mRA7b0JrpUvrA.jpeg" /></figure><p>One of the grand theories of our age is that spoilers are a cardinal sin. What utter hogwash. With <em>Caledonian Road </em>you know from the start how it’s all going to end. Celebrity writer cum academic cum high society figure Campbell Flynn’s life is going to end in tatters. Our pleasure as a reader lies not in the fact that this is withheld from us but that we don’t know <em>how </em>his downfall is going to come about. I enjoyed this book way too much to be honest. I hadn’t heard of Andrew O’Hagan before reading this book and purely bought it for the cover, only to be very pleasantly surprised by a novel of Dickensian proportions. Arguably, the fact that I’ve spent a) eight years in the UK at this point and b) far too much of this time reading about the ins and outs of London’s interconnected elites is a major contributing factor.</p><p><strong>Out of the Darkness. The Germans 1942–2022<br></strong><em>Frank Trentmann, 2023, Allen Lane</em></p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*v6HKFzQ8BJaqBbQijAJ5Hg.jpeg" /></figure><p>‘Out of the Darkness’ reached me as a gift by my wonderful DPhil supervisor Ralph — who thought that I would enjoy a bit of light reading of *checks notes* 700+ pages on German history after passing my viva (Ralph, if you are reading this, you couldn’t have given me a better present). Trentmann starts his book off by reminding us that he’s spent the better part of his life outside of Germany and that his ‘insider/outsider’ perspective proved instrumental in writing this book. It’s a view I can — for obvious reasons — identify with. After living in the UK for 8 years (and counting) I’ve come to see the country I come from in quite a different light; for good and ill. It has also (at least somewhat) removed the veil and allowed me to see many things more clearly — and this book has intensified that sensation. Trentmann takes us on a journey through recent German history but looks at it through the lens of moral struggle, illuminating how the German people have navigated feelings of conscience, compassion and complicity the last 100 years. And, in the author’s own words, the book ‘breaks with the success story with which we have become familiar […] [of a] teleological bent of progressing towards a better present’ instead showing that for Germany and its people there wasn’t and isn’t a single destination. Who we are as a people is complex and Trentman takes great care in foregrounding this, not least by showing how ‘contemporaries experienced their own times’, through the careful use of archival material. Despite this rather heavy sounding snapshot, ‘Out of the Darkness’ is surprisingly easy to read and eye-opening in a way that few other books are.</p><p><strong>Free Agents. How Evolution Gave Us Free Will<br></strong><em>Kevin J. Mitchell, 2023, Princeton University Press</em></p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*YQcEqrfnWUyxzEokhpQHpA.jpeg" /></figure><p>A warning ahead: This book is a rather challenging read. You will need focus and isn’t suited to hurried reading. Mitchell basically tries to address one tiny and simple question: <em>Free will, do we have it? </em>Drawing on a wealth of research across a variety of scientific and philosophical fields, he takes us on a journey from bacteria to artificial intelligence and back (read the book to understand to see what I mean) and questions how agency, or the ability to make decisions and act, originally developed. As I interpret it, his fundamental argument is that free will is a biological function that has evolved, reliant on neural capacities — essentially, he attributes agency to natural processes. Organisms, or so Mitchell argues, have evolved to act with purpose and for reasons (the very basic one being that we try to survive). And while we may not always be able to choose all of those purposes, we can reason about them and and exercise at least some self-control.</p><p>The book can be a bit too heavy on the detail at times. But at least no one can say that Mitchell does not show us the receipts or assumes that we as readers are idiots that can be hoodwinked with a bit of magical handwaving. Instead, he methodically navigates us through this complex topic. While some might find this overwhelming (to which I say, fair enough), for others like me, it works well. And if you end up not liking it, well, in that case you can exert some free will and put it aside.</p><p><strong>The MANIAC<br></strong><em>Benjamín Labatut, 2023, Pushkin Press</em></p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*JRfMZ17o-8UeiMnO1ulsjw.jpeg" /></figure><p>This series has previously included a biography of John von Neumann (see the 2023 edition). However, I could not resist including Labatut’s “The MANIAC” this year, too. I was unaware of his works until he gained widespread attention in the UK press earlier this year (who says that no one reads literary reviews any more?) What sets this book apart from Ananyo Bhattacharya ‘The Man from the Future’ is that it is not a conventional biography because it is both a fictionalised account of von Neuman’s life and narrated from the perspectives of others. To be honest, I found this quite the unsettling read and to me the book serves as a sobering reminder that any cult of genius is inherently a bad idea, regardless of the impressive accomplishments of the individual in question.</p><p><strong>AI Snake Oil. What Artificial Intelligence Can Do, What It Can’t and How to Tell the Difference<br></strong><em>Arvind Narayanan &amp; Sayash Kapoor, 2024, Princeton University Press</em></p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*afMtXBjxpHajhK8iNeuF3g.jpeg" /></figure><p>AI Snake Oil is a great book for anyone looking to gain a comprehensive understanding of artificial intelligence. Arvind and Sayash have done an admirable job of breaking down complex ideas, making them accessible and easy to grasp, which is no small feat especially on a topic so sprawling and complex. It’s a quick read, too, and devoid of the dense and overcomplicated prose that afflicts some other books on the topic. For those interested in catching up on fundamental discussions about AI, including its mechanisms, the implications of automation, and concerns regarding AI safety, this is a fantastic resource — and if you are one of those who work in the field but have relatives who go ‘Oh you mean killer robots’ every time you talk about your work, this is a good stocking filler…</p><p><strong>As Gods Among Men. A History of the Rich in the West<br></strong><em>Guido Alfani, 2024, Princeton University Press</em></p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*lWKgxZyY4Tl11ryH-DwOyg.jpeg" /></figure><p>Who reads books for their covers: me! Isn’t this one dazzling? So is the content which is…well…about the history and status of the rich in West, as the title says. On this one, I am going to be lazy and instead point you to a <a href="https://blogs.lse.ac.uk/lsereviewofbooks/2024/05/07/book-review-as-gods-among-men-a-history-of-the-rich-in-the-west-guido-alfani/">good review by Noah Sutter in the LSE Review of Books</a> because I a) haven’t read the final chapter yet but wanted to include the book nonetheless and b) he does a much better job at summarising what the book is about.Music</p><h3>Music</h3><p>No, <strong>Jacob Collier</strong> has still not made a Christmas album (hurry up!). In previous years, we got <a href="https://youtu.be/njcAgHWoS-8?si=zc4hutbfkHAyq6JA&amp;t=11">his version of ‘White Christmas</a>’, a (slightly too space-y) <a href="https://youtu.be/o9CusMKhoqk?si=_gMpG9tGv89aswxk">version of the ‘Christmas Song</a>’ and this rather lovely version of <a href="https://youtu.be/mrjyYZm0Dao?si=T96kUky5M-ccum7B">‘Have Yourself a Merry Little Christmas’</a> with Gregory Porter. This year, he’s given us three classics recorded with a (very involved) live audience at Abbey Road studios which are well worth a listen — even though he is admittedly not to everyone’s taste.</p><iframe src="https://cdn.embedly.com/widgets/media.html?src=https%3A%2F%2Fwww.youtube.com%2Fembed%2FojQI3clg40o%3Ffeature%3Doembed&amp;display_name=YouTube&amp;url=https%3A%2F%2Fwww.youtube.com%2Fwatch%3Fv%3DojQI3clg40o&amp;image=https%3A%2F%2Fi.ytimg.com%2Fvi%2FojQI3clg40o%2Fhqdefault.jpg&amp;type=text%2Fhtml&amp;schema=youtube" width="640" height="480" frameborder="0" scrolling="no"><a href="https://medium.com/media/630c689acd6adb7c04dc96444847f75b/href">https://medium.com/media/630c689acd6adb7c04dc96444847f75b/href</a></iframe><p>This next one is from last year but jazz pianist <strong>Christian Sands’</strong> ‘Christmas Stories’ is a very elegant and well-done holiday album of the ‘pour yourself a Whisky and fall asleep on the couch in front of the fire while he taps the keys in the background’–sort. Highly recommended.</p><iframe src="https://cdn.embedly.com/widgets/media.html?src=https%3A%2F%2Fwww.youtube.com%2Fembed%2Fvideoseries%3Flist%3DPLAym87g-oy7od3iBC-o5kPQuFYCRvgASh&amp;display_name=YouTube&amp;url=https%3A%2F%2Fwww.youtube.com%2Fplaylist%3Flist%3DPLAym87g-oy7od3iBC-o5kPQuFYCRvgASh%26si%3DesMsKUUC19GMRH2f&amp;image=https%3A%2F%2Fi.ytimg.com%2Fvi%2Fxc8b49571CY%2Fhqdefault.jpg%3Fsqp%3D-oaymwEXCOADEI4CSFryq4qpAwkIARUAAIhCGAE%3D%26rs%3DAOn4CLBg2mGXwkD5V-pNDyds3uilOKMylQ%26days_since_epoch%3D20060&amp;type=text%2Fhtml&amp;schema=youtube" width="853" height="480" frameborder="0" scrolling="no"><a href="https://medium.com/media/8e174b018c636ffe8fccfad4cee79e16/href">https://medium.com/media/8e174b018c636ffe8fccfad4cee79e16/href</a></iframe><p>Another one I rather enjoy, again from last year, comes from German jazz trumpeter <strong>Till Brönner</strong> with his album that does not ask but certainly answers the question ‘Can you play a trumpet so smoothly that it sounds like molten butter?’ Full playlist <a href="https://youtu.be/j1wEYEjCipk?si=GdoobnUfwTIp4DJN">behind the link</a> and on YouTube:</p><iframe src="https://cdn.embedly.com/widgets/media.html?src=https%3A%2F%2Fwww.youtube.com%2Fembed%2Fj1wEYEjCipk%3Ffeature%3Doembed&amp;display_name=YouTube&amp;url=https%3A%2F%2Fwww.youtube.com%2Fwatch%3Fv%3Dj1wEYEjCipk&amp;image=https%3A%2F%2Fi.ytimg.com%2Fvi%2Fj1wEYEjCipk%2Fhqdefault.jpg&amp;type=text%2Fhtml&amp;schema=youtube" width="640" height="480" frameborder="0" scrolling="no"><a href="https://medium.com/media/3568f9d1f68fe59b21f5012a8091754b/href">https://medium.com/media/3568f9d1f68fe59b21f5012a8091754b/href</a></iframe><p>Finally, <strong>Joy Lapps and Jarnell Lewis</strong> have recorded the Carribean Christmas Mixtape — which has to be experienced. Again, the <a href="https://youtube.com/playlist?list=PLP4aBNUPvLhYomlAg0b9HY7LhTOUOAy4d&amp;si=9WsAys4UgdexQKbq">full playlist is here</a> and below.</p><iframe src="https://cdn.embedly.com/widgets/media.html?src=https%3A%2F%2Fwww.youtube.com%2Fembed%2Fvideoseries%3Flist%3DPLP4aBNUPvLhYomlAg0b9HY7LhTOUOAy4d&amp;display_name=YouTube&amp;url=https%3A%2F%2Fwww.youtube.com%2Fplaylist%3Flist%3DPLP4aBNUPvLhYomlAg0b9HY7LhTOUOAy4d%26si%3D9WsAys4UgdexQKbq&amp;image=https%3A%2F%2Fi.ytimg.com%2Fpl_c%2FPLP4aBNUPvLhYomlAg0b9HY7LhTOUOAy4d%2Fstudio_square_thumbnail.jpg%3Fsqp%3DCL6p8LoG-oaymwEICKoDEPABSFqi85f_AwYI4rnbuQY%3D%26rs%3DAOn4CLAFYwxNxU0P2xwBlHraPLZZQbhR7g%26days_since_epoch%3D20070&amp;type=text%2Fhtml&amp;schema=youtube" width="853" height="480" frameborder="0" scrolling="no"><a href="https://medium.com/media/8fe829e85d205b88ed60dd65e745b114/href">https://medium.com/media/8fe829e85d205b88ed60dd65e745b114/href</a></iframe><p><em>Dr</em> <a href="https://www.felixsimon.net/"><em>Felix Simon</em></a><em> is a Research Fellow in AI and News at the Reuters Institute for the Study of Journalism and a Research Associate at the Oxford Internet Institute, both at the University of Oxford. He is on </em><a href="https://bsky.app/profile/felixsimon.bsky.social"><em>BlueSky</em></a><em>.</em></p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=cd6bec59b96c" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Acknowledgments]]></title>
            <link>https://felixsimon.medium.com/acknowledgments-98e722a86477?source=rss-41832dfb135a------2</link>
            <guid isPermaLink="false">https://medium.com/p/98e722a86477</guid>
            <dc:creator><![CDATA[Felix M. Simon]]></dc:creator>
            <pubDate>Wed, 12 Jun 2024 15:23:40 GMT</pubDate>
            <atom:updated>2024-06-17T15:43:01.314Z</atom:updated>
            <content:encoded><![CDATA[<h4>Maybe the real DPhil was the friends we made along the way</h4><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*s7xwGbc1erRFSwy6fDyuog.jpeg" /><figcaption>Source: Map of Days, Grayson Perry — Balliol College, Oxford</figcaption></figure><p><em>Maybe the real DPhil was the friends we made along the way — </em>Internet Proverb</p><p>I always thought that acknowledgments are among the most interesting pieces of writing. Usually, they are the first thing I read in any book or paper. Why, I cannot quite say. I assume it betrays a fascination with the traces we leave in each other’s lives and works. Nowhere is this usually clearer acknowledged then in the section named after this particular act where people meet who have never (and will never) encounter each other. But despite their seeming simplicity (after all, what can be so difficult about listing some names and saying ‘thanks’ you, dear reader, might say?), acknowledgments I found them to be surprisingly hard to write, mostly because the possibilities seem endless: Where to start? What and who to leave in — and out? <a href="#_ftn1">[1]</a> And how to find the right balance with a genre that always teeters on the brink of too much detail and pathos (but consider this: when else do you get the chance to be effusive in your thanks and grateful to the many people that have made you and your work possible in myriad visible and invisible ways? I daresay the opportunities are limited). Perhaps, for all these reasons they are the last form of writing that should be automated by AI, even though their very formulaicness yields itself perfectly to the same. Mind you, I tried, for fun, but found the result lacking and soulless. In their best possible form, acknowledgments are written from a personal place, imbued with feeling, and drawing from a well of experience that is beyond the reach of any computer system, no matter how sophisticated. One final thought is worth mentioning in this brief preamble: In many ways, what I like about acknowledgments is that they are the best rebuttal to the idea of the ‘lone genius’ that we still hold so dear collectively. So let me use this space to call this bluff in my case. That my DPhil was possible at all was only because I had support — because people saw things in me that I failed to see, convinced me that I could do things I doubted I could do and otherwise carried, inspired, and motivated me. Now, having bored you long enough, who are these people?</p><p>First, I would like to acknowledge the support of the Leverhulme Trust, the Minderoo-Oxford Challenge Fund and TORCH, the Dieter Schwarz Foundation, Columbia University’s Tow Center, Balliol College, the Oxford Internet Institute, and the University of Oxford Covid-19 Scholarship Extension Fund who variously provided funding for different aspects of this and related projects. My DPhil would not have been possible without their generous support. I should also give a brief shoutout to the city of Oxford and its people, but especially some of the places where I wrote, debated, and defended this thesis: Najar’s, The Lamb and Flag, the Gardener’s Arms, the Perch, the Victoria, Common Ground and Society cafes, as well as the Bodleian Libraries. <a href="#_ftn2">[2]</a></p><p>My second round of thanks goes out to the ‘Leverhulmians’ as I like to refer to them in my imagination: the members of the <em>Leverhulme Doctoral Centre</em> ‘Publication Beyond Print’ here at Oxford University, but especially Daniel Wakelin, Daria Kondakova, Matthew Ball, Barbora Sojkova, Eddie Jones, Natascha Domeisen, Katie Noble and Aoife Ní Chroidheáin. Each of them provided crucial moral and intellectual support on this long journey and opened up new intellectual worlds to me. Who would have thought that you can find links between samizdat literature, Roman coins, and artificial intelligence in publishing? Certainly not me, but I am glad we went on this adventure together.</p><p>At the <em>Oxford Internet Institute</em>, I owe more than I can say to Maisy Gills, Karen Mead and Chrissy Bunyan — I hope the institute never forgets the assets they have in you and the place would not be the same without you. Thank you for making me and M feel at home at this department. Marten Krijgsman saved me more than once from administrative purgatory. Ornella Scutio not only always had an open ear for my concerns and worries but also crucially taught me how to negotiate and — together with Duncan Passey, Carolyn Watson, Lucy Hennings and Lindsay Campbell — how to navigate the maze that is university HR and the world of research grants. To Sara Spinks, Roz Pacey, Mark Malbas and Ellen Mobbs I owe the promotion of my work and the chance to build a profile for myself in ways that few PhD students can hope for. I thank the IT department for finally teaching me how to use the printers four years into my DPhil. My former assessors Grant Blank and Jonathan Bright shaped my DPhil project in the best possible way by providing straightforward, honest and actionable feedback. As I have learned from Henry Farrell and Hugo Mercier, ‘serious criticism is probably the most valuable contribution we can make to the cognitive division of labour’ <a href="#_ftn3">[3]</a> and both (as well as Seth Lewis who will find himself mentioned here again) have granted me the kindness to engage with my work in a serious and honest ways — and it has been all the better for it. There is also a long list of OII people past and present who have played their part in my DPhil journey (sometimes without realising), so in alphabetical order I would like to thank Kathryn Eccles, Lily Davies, Monah Elswah, Rebecca Enyon, Sophie Cray, Carl-Benedikt Frey, Phil Howard, Adam Mahdi, Nahema Marchal, Helen Margetts, James Manyika, Adam Mahdi, Jakob Mökander, Lisa-Marie Neudert, Julia Lefkowitz, Nahema Marchal, Lulu Shi, David Tomchak, Oleh Stupak, Rosaria Taddeo, Greg Taylor, and Pu Yan. I am also grateful to the rotating cast of MSc students who suffered through my teaching, whose rooms I occupied at 1 St Giles all too frequently (especially on the home stretch of my thesis), and whose company and friendship I enjoy(ed), especially Hannah Stewart, Aurora Hawcroft, Boxi Wu, Elijiah Lewien, Emmie Hine, Frida Shui, Gill Whitehead, Hannah Perry, Ingrid Epure, Jake Stein, Joey D’Urso, Justin Yeung, Karin Vatja Engström, Marlene Straub, Megha Mishra, and Susanne Klausing. It is wondrous to see what you are up to, and I will follow your careers with great interest. Finally, I owe many thanks to Vicky Nash for graciously given advice and for being the first to advise me on the idea to pursue a PhD during my master’s degree. Her counsel to always think about the opportunity costs I have carried forward in more than one way. Andrew ‘Andy’ Przybylski became a crucial mentor on my way through more than one academic and personal minefield. He taught me not only to never be satisfied with easy answers, but also how to resist the hype and that integrity is a choice we can and have to make over and over again. Gemma Newlands took me under her wing as soon as she arrived at the OII and played a crucial role in providing me with honest advice about academic life, and lots of banter to sweeten the medicine — I will never not be impressed by her rigour, wit, and work ethic. Within my wider cohort of fellow DPhil students, I am thankful to Cassidy Bereskin, Paul Röttger, Tomas Borsa, Roland Chen, Kobi Hackenburg, Thomas Hakman, Prathm Juneja, Jakob Mökander, Huw Roberts, Lily Rodel, Martha Ziosi, Clementine Collett and Cailean Osborne for their advice and being such wonderful company.</p><p>My college <em>Balliol</em> provided not just an academic home, but also a home away from home. The Balliol MCR was a source of many enjoyable events and intellectual stimuli and Alex Russell, Carol Jones, and Konrad Ksiazek are the main reasons that I had so much fun in this community. The college fellows — Dame Helen Gosh, James Forder, Martin Burton, John Tasioulas, John-Paul Ghobrial and especially Nicki Trott and Melanie Marshall — and college staff — in particular Bruce Rossiter, Bélen Campos Romero, Jordan Moore, Michael Hooker, Chris Puddifoot, Greg Barker, and Bertrand Faucheux — provided for my comfort and academic success in many ways, from helping with administrative matters and providing advice, to help with events, unwieldy ancient doors (thanks again Martin!), dietary matters, and so much more. Meanwhile, my ‘Balliol Banter Crew’ — Nick Dickinson, Linda Eggert, Bysshe Inigo Coffey, Max Marcus, Lauren Burgeno, Sophie Marnette, Freddy Smith, Jim Mallinson, Richard Norman and Raghavendra Srinivas — not only accepted me as the honorary graduate member but also showed me a side of college life that I had no idea existed. Thank you all, it was an honour.</p><p>At the <em>Reuters Institute for the Study of Journalism</em> I am grateful to Kate Hanneford-Smith, Nic Newman, Mitali Mukherjee, Federica Cherubini, Caithlin Mercer, Matthew Leake, Louise Allock, Rebecca Edwards, and Eduardo Suárez for various forms of support and encouragement. Kirsten Eddy, Camila Mont’Alverne, Amy Ross Arguedas, Benjamin Toff, Craig T. Robertson, and Waqas Ejaz provided feedback on my work and former visiting fellows Jessica Cecil, Seth Lewis, and Alfred Hermida collaborated with me on projects that proved to be very helpful to my DPhil. Sílvia Majó-Vázquez believed in me when I did not. Her reference letter might have been the one that made all the difference in getting me into my programme. Finally, Rasmus Nielsen and Richard Fletcher provided ample feedback on my work at various points and gave me not only continuous employment as a research assistant but also the chance to learn from some of the brightest minds in communication research which made me a much, much better scholar. My time at Oxford would have been lesser without you all and I would not have been able to become the academic I am today without the support of the institute.</p><p>The <em>community of academics</em> that has inspired and supported my work is far too large to name them all, so I will leave it at some special mentions, with a big thank you to my former lecturers Julika Griem, Rembert Hüser, Sophie Einwächter and Vinzenz Hediger. The latter also become a co-author and collaborator and joins the company of Annika Sehl, Scott Babwah Brennen, Hugo Mercier, Sacha Altay, Chico Camargo, Luisa Fernada Isaza Ibarra, Alexandra Borchardt, Valeria Resendez Gómez, Kim Björn Becker, Seth Lewis, Chris Crum, Benjamin Toff, and Katherine Dunn who worked with me on various projects during my DPhil. Andreas Jungherr always had advice for me and Valérie Bélair-Gagnon and Sarah Anne Ganter both made time for me when I got stuck. Natali Helberger and Claes de Vreese showed understanding at a crucial juncture while Theresa Josephine Seipp and Tomás Dodds discussed important points of this thesis with me — and became friends in the process. The International Journalism Festival and Arianna, Chris, and Francesca allowed me to return and present my work three times while my Twitter community stepped to the fore more than once in helping me with feedback and advice and otherwise entertaining my all-too-frequent musings. Finally, I am incredibly grateful to my viva assessors Carl-Benedikt Frey and Caitlin Petre — two academics I tremendously admire and who took time out of their busy schedules to read and assess my doctoral work.</p><p>No acknowledgments, however, would be complete without the many <em>friends</em> that I made (or who stayed with me) during my degree. Uta Hillebrand has been there for me all these years, and so have Gerdi Herold-Ehlerding, Karin Fleck, Matthias Kuhn, Dominique Zilch and Florian Gerth, Hannah Windau and Matthias Kemmerer, Thomas Reynaert and Stefanie Hatzfeld, as well as Barbara Demedts and Gary James. From my masters’ days, Eddie Percarpio, Adena Meriel, Olivia Gonzalez, Aiste Zalepuga, Anisha Nadkarni, Thomas Vogl, and Robert Gorwa were part of this journey, even from a great distance. Jan Rau helped me through more than one mental health crisis and his been a loyal friend and advisor ever since I first met him. From Elisabeth Gamperl I have learned what fighting spirit means and that hope is to be found in the darkest places, even if you are going through literal hell. William Allen and Hannah Forde selflessly helped me with more advice than I could have ever asked for. With Siddharth ‘Sid’ Venkataramakrishnan and Sabrina Ahmad I have gone through thick and thin, and they are better friends than I deserve them. Maria Farrell taught me about the duty of hope and together with Ed Wingfield, Milo and Molly not only welcomed me and M into their London community and home but also provided some much-needed perspective about what truly matters in life. Mark Lee became my wise owl at the halfway point, always ready to chip in with reading suggestions and career advice. I hope to still have the same curiosity and open-mindedness once I have reached his level. Emily Bell, I had the good fortune to meet at around the same time and have cherished her advice and perspective, as well as her friendship, ever since. Julie Posetti (with Tim, Amalia and Toffee) and Nabeelah Shabbir knew me way before I ended up back at the OII and have continued to support me throughout my time at Oxford — with job and housing advice, places to stay, good food and much more. Kim Björn Becker and Tomás Dodds are not only two of the most reliable and diligent co-authors I could have asked for but also turned out to be two wonderfully warm people, with good wit, intellect, and magnanimity in equal measure, who have both done much to help me along. Seth Lewis not only shaped my thesis in central ways through his feedback, he is also a true mensch and exemplifies what a scholar should be: he combines a sharp intellect with an extraordinary work ethic, and his kindness and compassion are boundless. His impact on both my academic work and personal growth has been profound, and I am glad we had the chance to meet before the Covid lockdowns ripped all our lives apart. Alexandra Borchardt is one of the handful of people who can all lay claim to getting me onto this trajectory and into this DPhil in the first place. Her steadfast support for me and her boundless optimism were crucial, and I could not have wished for a better mentor, friend, and co-author than her. Likewise, Erik Bucy not only provided level-headed advice and friendship at the very beginning but also ever since.</p><p>In Oxford, Aoife Ní Chroidheáin, Trisha Prahbu, Peter Andringa, Linda Ma, Kate Joyness-Burgess, Laura Herman, Patrick Gildersleve, Yayoi Teramoto, and Nathan Lane all deserve special mention as fellow travellers and friends. Nick Dickinson and Linda Eggert graciously overlooked the fact that I was still a graduate student and made me and M part of their community, while inspiring me with their seemingly inexhaustible knowledge of all things philosophy and politics. They are also the beginning of the chain that led me and us to Richard Norman, James Fletcher and George Balkwill who all became dear friends and together provided much needed distraction from the chores of academic life, as well as showing us a different side of university life. Mary Sanford cracked my shell very early on and was my first new friend in my DPhil cohort. Wise beyond her years and funnier than she will be willing to admit, Mary helped me transition back into Oxford life after an absence of two years and has continued to be a friend and source of inspiration ever since. Anne Ploin provided crucial early advice, banter, and coffee vouchers from which I still draw. Charlie Smith showed me that you can be a hard-working scholar while still having fun at it. I might have failed to adopt his productivity hacks but have learned countless other things from him in the process. Liam Bekirsky taught me that ‘in the kingdom of the nerds, even a nerd can be king’ and how to take life less seriously (in the best possible way). Claire Leibowicz became my ‘AI buddy’ at a distance and has championed me in ways that I probably didn’t deserve and inspired me in more ways than she knows. Thanks to Amanda Curtis I did not only find my love for matcha but also learned more about #firstcousins than was good for me. Someone with a better sense of humour is hard to find. Jaimie Lee Freeman kept me sane in the side chat throughout countless Teams meetings and always had time for my worries while bringing me some much-needed distraction. Hubert Au sharpened my knives — both figuratively and literally — and was never afraid to talk about the hard things, and together with Hayyu (Inda) Imanda wined, dined, and entertained me and M on many wonderful occasions. Joanna Rivera-Carlisle I thank for the Austrian banter, being a general bringer of good times and the most joyous laugh in the department. Paul Rivera-Carlisle I am grateful to for the best drinks in the whole of Oxford, but most importantly for bonding over our mutual love of imitating Werner Herzog. We have looked ‘ze chicken’ in the eye more than once.<em> </em>Who knew that the abyss could be such a fun place? It truly was an experience bordering on ‘ze sacred’. Nayana Prakash was my fellow traveller and confidante to the very end of the DPhil process. I will miss her community spirit and those writing days in the library and our lunches in college when we discussed gossip and Harry Potter while we were supposed to study. Hannah Kirk, I have only met properly far too late into my DPhil journey but am so thankful that I did. She has endured far too many of my stupid questions about LLMs, went full goblin-mode with me in the OII dungeons to make sure that we finished those papers, and trusted me with her story on more than one occasion which I see as the highest honour. Chris Crum became a role model in the way he approached his DPhil and a good friend, and later co-author, too. How someone could balance a law degree, a PhD, rowing, and a social life at the same time is still beyond me and I probably could have finished my own degree two years early if I had had just a shred of the same quality. Needless to say, having him as an inspiration likely prevented me from taking even longer. Benjamin Toff I first met as a master’s student and then again many years later during my doctoral degree. His advice during the final stages of my DPhil helped me in navigating a difficult academic job market, his academic work deeply impresses me, and his friendship is one that I hope to enjoy for many years to come. Chico Camargo is not only a true Renaissance man but also the academic and human I wish I could be. Endlessly curious and gifted, he was one of the first to demystify AI for me and continues to be an inspiration in whatever I do. His life advice allowed me to avoid many bad decisions over the last three years. Maggie Mustaklem (including Sam, Teddy, Bianca Amal and Arnie) and Isabel Ebert — my ‘Brain Freeze’ group — were the steady rhythm of my DPhil experience. On many days I woke up with and went to bed with your messages and I have learned much from you both as an academic and as a human and this thesis would be in worse shape without your help. To Lily and Christian, I am grateful for their continued friendship and for adventures during my DPhil that I will remember for a lifetime. I never expected to write the words ‘It was fun to get vaccinated together’ but here we are. Not a day goes by where I don’t wish that you lived closer. The same is true for Kirsten and Bryan Eddy and Camila Mont’Alverne whom I miss dearly and who form(ed) part of my and M’s Oxford support bubble together with Amy Ross Arguedas and Dan Patino. Our ‘Formal Group’ not only gave and gives me a steady supply of memes, good vibes and a lifetime of fun memories, but also academic counsel aplenty. All the other important things I have already told you all individually but let me just reiterate here that Oxford has never quite been the same again without you. Finally, Verena Gautsch made my and M’s Oxford years better in so many ways. We both could not have wished for a better friend during our time here (and beyond) and I am immensely glad she chose Oxford as her place of study, despite her grim first impression for which I take full blame.</p><p>At last, I would like to thank my <em>family</em> for making me the person I am today — for good and ill — and accompanying me on this journey with love and encouragement. My sisters Carla and Sophia for their love which I have failed to return in kind on more than one occasion. Both are taking the world in their stride and constantly remind me of experiences and minds beyond the reach and comprehension of my own, each in their own way. Susi and Klaus König and Gabi Wendtner for many joyful hours of distraction. Monika-Schmid Wendtner and Clemens Wendtner for crucial health advice and support and allowing us to escape ‘the Boris experiment’ just in time. To my cousin Lukas König who is more accurately described as a brother to me and whose companionship I cherish. To my grandparents Renate and Hans Wendtner, for instilling in me the sense that wanting to know is not a crime and teaching me one trick or the other on how to survive in the world. And to my grandparents Renate and Hans Simon for their unwavering support and encouragement and reminding me that I am more than my work. And to both of them for the values and work ethic they have taught me over many years. I may have failed to become a medical doctor, but I hope this doctorate nonetheless will make you all proud — and I am glad that you are here to see this long journey reach its conclusion. Finally, there are two sets of parents to which I owe my deepest thanks. Heike und Stephan Disser have nurtured and supported me at every step along the way, welcoming me into their family and home with open arms. Especially during the unsettling times of the pandemic, they provided me with a sanctuary that allowed me to continue my work. Your warmth and unwavering support have been a beacon of light in the most challenging times. Finally, to my parents Dagmar and Dieter. With your stoic dedication, you have provided the support and counsel that have been the bedrock of my intellectual adventures ever since I first annoyed you with questions about nuclear power plants as a two-year-old in the bathtub (admittedly, I have strayed from this, subject-wise). You did not give up on me when my academic record suggested less than stellar prospects and stood by me when certain teachers seemed convinced that I was just too lazy and not bright enough for anything in life (to whom I can only say in this brief moment of winning, ‘well, well, well, who’s laughing now?’). Your never-ending patience for my various follies fostered a resilience and a quiet confidence in going my own way that I’ve carried into every chapter of my life. For your wisdom, your pragmatism, love, and the subtle ways you’ve championed my pursuits, I extend my deepest thanks. You have both shaped my world in profound yet unspoken ways.</p><p><em>The</em> <em>most important part</em> I have saved for the very end. First, I am immensely thankful to all my participants whom I cannot name here but without whom this thesis simply would not exist — for your time and willingness to talk to me I cannot thank you enough. I hope this work does you justice. These acknowledgements would also not be complete without a special tribute to two key pillars of this thesis: my two supervisors, Gina Neff and Ekaterina Hertog. It was Gina who set me on this path; her belief in AI and news as a subject for research was visionary. Long before it was a topic of wide interest, she recognised the potential and was instrumental in securing the initial funding for my DPhil. Her decision to back me, a choice that I reflect on with a blend of wonder and gratitude, was a defining moment in my academic pursuit. To this day I am not sure what moved her to think that I’d be the right candidate for this work, but I am eternally grateful that she put her trust in me and took me on as her student. Ekaterina ‘Katya’ Hertog saw me through the final stages of my doctoral work. At times when the finish line seemed elusive, Katya’s timely and pragmatic support kept the wheels turning. Her readiness to provide support at a moment’s notice, regardless of the hour, played an invaluable role in not only the finalisation of my DPhil but also in preparing me for future academic ventures. The debt of gratitude I owe for her supervision, always delivered with warmth and at all hours, is significant. Such commitment is rare, and it is something I aspire to emulate and pay forward in my future endeavours.</p><p>There are two people who deserve my thanks more than anyone else. One of them is my main supervisor, co-author, and spiritus rector Ralph Schroeder. If any doctoral adviser was like Ralph, the world of academia would be a much better place. No one during my academic journey has shown so much vision, worked harder on my behalf or provided better counsel than him. Unfazed and undaunted by even the most complex situations, his generosity, selflessness, and desire to follow knowledge like a sinking star, to quote Tennyson, will always remain an inspiration to me. He saw hope where I did not and has shaped my thinking in more ways than I can list here — including on the benefits of walking to solve complex puzzles. His and Jen’s support have meant the world to me, and I am glad that I can still call him a friend at the close of this chapter, considering that I often did not make life easy for him.</p><p>Finally, to Michelle — my fiercest critic and most ardent advocate and supporter. There are no words in any language known to humankind that could do your contribution to this thesis and my life justice and I can have no hope of ever expressing how much I owe to you. Any reputation I might have for intellectual prowess comes, in truth, from the relentless daily engagement with a mind far surpassing my own in wit, sharpness, and agility — and the fact that I am hopefully not an utter failure at being a normal and tolerable human being, I have above all you to thank for. Knowing better than anyone your thoughts on my penchant for flowery language, I shall leave it at that, confident in the knowledge that you will find meaning in what is left unsaid here. Onwards and upwards 🥊</p><p><a href="#_ftnref1">[1]</a> To all those who did not find their names here: Bear with me, these acknowledgments were written in a rush, and you know where to complain.</p><p><a href="#_ftnref2">[2]</a> Let no one ever say acknowledgments cannot be useful, e.g. as a travel guide.</p><p><a href="#_ftnref3">[3]</a> <a href="https://crookedtimber.org/2020/07/24/in-praise-of-negativity/">https://crookedtimber.org/2020/07/24/in-praise-of-negativity/</a></p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*7D8_5YnRLsG37YYMw8cKCw.png" /><figcaption>How DALL:E imagines my DPhil work (in the style of Gustav Klimt)</figcaption></figure><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=98e722a86477" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[AI Search & Implications for Publishers and the Public Sphere]]></title>
            <link>https://felixsimon.medium.com/ai-search-implications-for-publishers-and-the-public-sphere-db931b6c8274?source=rss-41832dfb135a------2</link>
            <guid isPermaLink="false">https://medium.com/p/db931b6c8274</guid>
            <category><![CDATA[journalism]]></category>
            <category><![CDATA[news]]></category>
            <category><![CDATA[ai]]></category>
            <category><![CDATA[technology]]></category>
            <dc:creator><![CDATA[Felix M. Simon]]></dc:creator>
            <pubDate>Thu, 16 May 2024 14:07:17 GMT</pubDate>
            <atom:updated>2024-05-16T14:07:17.813Z</atom:updated>
            <content:encoded><![CDATA[<h3>AI Search &amp; Some Implications for Publishers and the Public Sphere</h3><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*5UciI_NwF14jRRoKyk8h_g.jpeg" /><figcaption>Yes, this is DALL:E generated</figcaption></figure><p>It’s only Thursday and it has already been a big week for AI announcements. I haven’t quite made up my mind about it, but here are some initial thoughts about the implications for tech companies, publishers, and the public.</p><p>Both Google and OpenAI seem to double down on “agentic” AI and multimodality. Seamless voice interaction and search is not new, of course, but if the demoed near-natural conversations without a significant lag prove to be reliable, interesting times could be ahead.</p><p>I am still not super convinced that voice interaction is the big flex tech companies think it is. People seem to generally like to consume written information (higher information density, easier to store/re-use, better privacy protection), but it might make a dent in traditional search and the way we interact with information.</p><p>More interesting to me is how Google is doubling down on bringing AI to search. It’s something many expected for months, and now it’s happening. Davey Alba and Julia Love have a good summary over at Bloomberg <a href="https://www.bloomberg.com/news/articles/2024-05-14/google-infuses-search-with-ai-in-a-fully-revamped-experience">here</a> or you can get it directly from <a href="https://blog.google/products/search/generative-ai-google-search-may-2024/">Google’s blog</a>.</p><p>For now, we don’t know for which searches SGE will kick in. Industry analyst Bart Horalewicz has some interesting data on this, suggesting that so far a majority of searches do not trigger SGE. But who knows if it will stay that way.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*Wn1nPUtoxlX_2zfAn7kRBQ.png" /><figcaption>Source: <a href="https://x.com/bart_goralewicz/status/1790438632081240095">https://x.com/bart_goralewicz/status/1790438632081240095</a></figcaption></figure><p>The <a href="https://apnews.com/article/google-search-ai-overviews-internet-traffic-ebb6bbbde17ed29a5f7b630d9e5e285b?mc_cid=ea3abe4265&amp;mc_eid=15fdf522f7">Associated Press</a> quotes an industry expert as saying that about 25% of traffic to publishers could be negatively affected by the de-emphasis on website links, with knock-on effects on ad revenue and subscriber numbers. All this comes against the backdrop of already declining referral traffic from platforms. As Press Gazette documents (h/t to Peter Erdelyi’s newsletter), it has gone down for both Google and Facebook. See their reports on <a href="https://pressgazette.co.uk/media-audience-and-business-data/media_metrics/facebooks-referral-traffic-for-publishers-down-50-in-12-months">Facebook’s referral traffic</a> and the <a href="https://pressgazette.co.uk/media-audience-and-business-data/first-google-core-update-of-2024-brings-bad-news-for-most-news-publishers">Google Core Update</a>.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*Zhu2lvkhhogQeYeQcltckg.png" /><figcaption>Not quite freefall but also not a case of “nothing to see here”. Source: PressGazzette, <a href="https://pressgazette.co.uk/author/aisha-majid/">Aisha Majid</a></figcaption></figure><p>With respect to their AI search SGE, Google claims that “we see that the links included in AI Overviews get more clicks than if the page had appeared as a traditional web listing for that query.” We have to take that at face value for now. Whether that will be borne out by the evidence is another question. The claim, of course, is that they will keep on sending traffic to publishers, but who knows how much. And they are not saying anywhere that traffic won’t decline (as far as I can tell).</p><p>The <em>question for publishers</em> is, of course, <em>how they can ensure that people will still 1) find and 2) consume their content (and for which reasons and needs) in an environment where many people already find many of their information needs covered elsewhere</em>. Reed Albergotti writing for Semafor put it more bluntly: “Publishers […] need to make tough decisions. How can they build businesses that they alone control?” How can they, indeed?</p><p>No matter how much publishers and critics wail, that’s of little help if users come to like this technology (and here being the de-facto monopolist over search helps a lot. Standard-setting power for the win…). As I put it to <a href="https://www.technologyreview.com/2024/05/14/1092407/googles-astra-is-its-first-ai-for-everything-agent/">Melissa Heikkilä and MIT Tech Review</a>: “This will help speed up the search process and get users more specific answers to more complex, niche questions […] I think that’s where Search has always struggled.” And answers don’t have to be perfect, just good enough in many cases. See Kevin Roose for NYT: “It may also be a popular feature with users — I’ve been testing AI overviews […] and have generally found it to be useful and accurate.” (<a href="https://www.nytimes.com/2024/05/14/technology/google-ai-answers.html">Full article here</a>).</p><p>And the question about the training data and copyright remains. Where does the training data come from? Who should be compensated and how? We never hear much about this in these announcements — instead, we get lots of safety and brave new world talk. This is also a task for regulators who have basically let this happen because of “innovation.” At least the House of Lords Digital Committee didn’t mince words in <a href="https://publications.parliament.uk/pa/ld5804/ldselect/ldcomm/54/5402.htm">its latest report</a>:</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/0*0u1Jl3JRpLy20XaL" /></figure><h3>The big picture</h3><p>It’s another step-change in the competition for audiences and attention. I think we will see a further tethering of people more closely to information and to these companies as providers of information (including news), with platforms taking over even more of the informational needs that people used to get through, e.g., news or other content on the web.</p><p>By giving users summaries instead of directing them to the original sources, it will likely take away the attention and money-making opportunities for those who create and invest in producing original information (such as publishers)… which wouldn’t be great.</p><p>Not only could we see a “flattening” and homogenization in terms of what users get to consume (one “average” answer instead of a diversity of results) by virtue of how these systems work; if the input becomes worse, it reinforces the issue. Political Scientist Andreas Jungherr has described this nicely over on <a href="https://x.com/ajungherr/status/1790647442511118578">Twitter/X</a>: it poses a threat to the idiosyncrasies of signals in the public arena.</p><p>As I put it in my recent <a href="https://www.cjr.org/tow_center_reports/artificial-intelligence-in-the-news.php">Tow Center report</a>, all this ultimately provides a pathway for tech companies to further cement their control over information, and potentially enabling them to take over tasks that were once central to, e.g., journalism. Control here to me means: the ability to (1) shape the structural conditions of the public arena (Who gets to see what? Who gets heard? Who gets to organize information?) in (2) the service of economic aims (who gets to benefit?) and connected to (3) political aims (which are often about shaping regulation to entrench and extend one’s control and position).</p><p>And the race to consolidate and extend this control is heating up.</p><p><em>(Last updated 16 May 2024)</em></p><p><em>Felix M. Simon is a Research Fellow in AI and Digital News at the Reuters Institute for the Study of Journalism at the University of Oxford.</em></p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=db931b6c8274" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[AI, news, & the dynamics of control in the media ecosystem]]></title>
            <link>https://felixsimon.medium.com/ai-news-the-dynamics-of-control-in-the-media-ecosystem-691053180d68?source=rss-41832dfb135a------2</link>
            <guid isPermaLink="false">https://medium.com/p/691053180d68</guid>
            <dc:creator><![CDATA[Felix M. Simon]]></dc:creator>
            <pubDate>Sun, 07 Apr 2024 11:31:28 GMT</pubDate>
            <atom:updated>2024-05-17T08:00:14.645Z</atom:updated>
            <content:encoded><![CDATA[<h4>Some reading recommendations</h4><figure><img alt="" src="https://cdn-images-1.medium.com/max/1012/1*7AsZhQAn06hxupyawl46jg.png" /></figure><h3>Reading recommendations</h3><p>Here some readings I found useful on this topic. To make it easier to navigate these, I have grouped them by topic.</p><h4>Control</h4><ul><li>Beniger, J. R. (1986). The Control Revolution: Technological and Economic Origins of the Information Society. Harvard University Press.</li></ul><h4>Large language models and generative AI</h4><ul><li><em>House of Lords Communications and Digital Committee: Large language models and generative AI</em> (1st Report of Session 2023–24; House of Lords Communications and Digital Committee). (2024). House of Lords. <a href="https://publications.parliament.uk/pa/ld5804/ldselect/ldcomm/54/54.pdf">https://publications.parliament.uk/pa/ld5804/ldselect/ldcomm/54/54.pdf</a></li><li>Wolfram, S. (2023, February 14). <em>What Is ChatGPT Doing … and Why Does It Work?</em> Stephen Wolfram. <a href="https://writings.stephenwolfram.com/2023/02/what-is-chatgpt-doing-and-why-does-it-work/">https://writings.stephenwolfram.com/2023/02/what-is-chatgpt-doing-and-why-does-it-work/</a></li></ul><h4>AI and the technology industry</h4><ul><li>Metz, C., Kang, C., Frenkel, S., Thompson, S. A., &amp; Grant, N. (2024, April 6). How tech giants cut corners to harvest data for A.I. The New York Times. <a href="https://www.nytimes.com/2024/04/06/technology/tech-giants-harvest-data-artificial-intelligence.html?unlocked_article_code=1.ik0.RaCf.bqzjdn7Qj4Gg&amp;smid=nytcore-ios-share&amp;referringSource=articleShare&amp;ugrp=u&amp;sgrp=c-cb">https://www.nytimes.com/2024/04/06/technology/tech-giants-harvest-data-artificial-intelligence.html</a></li><li>Vipra, J., &amp; Myers West, S. (2023). <em>Computational Power and AI</em> (p. 32). AI Now Institute. <a href="https://ainowinstitute.org/wp-content/uploads/2023/09/AI-Now_Computational-Power-an-AI.pdf">https://ainowinstitute.org/wp-content/uploads/2023/09/AI-Now_Computational-Power-an-AI.pdf</a></li><li>Ahmed, N., Wahed, M., &amp; Thompson, N. C. (2023). The growing influence of industry in AI research. <em>Science</em>, <em>379</em>(6635), 884–886. <a href="https://doi.org/10.1126/science.ade2420">https://doi.org/10.1126/science.ade2420</a></li></ul><h4>AI, journalism, and technology companies</h4><ul><li>Simon, F. M. (2022). Uneasy Bedfellows: AI in the News, Platform Companies and the Issue of Journalistic Autonomy. <em>Digital Journalism</em>, <em>10</em>(10), 1823–1854. <a href="https://doi.org/10.1080/21670811.2022.2063150">https://doi.org/10.1080/21670811.2022.2063150</a></li><li>Simon, F. M. (2023). Escape Me If You Can: How AI Reshapes News Organisations’ Dependency on Platform Companies. <em>Digital Journalism</em>, <em>12</em>(2), 149–170. <a href="https://doi.org/10.1080/21670811.2023.2287464">https://doi.org/10.1080/21670811.2023.2287464</a></li><li>Papa, V., &amp; Kouros, T. (2023). Do Facebook and Google Care about Journalism? Mapping the Relationship between Affordances of GNI and FJP Tools and Journalistic Norms. <em>Digital Journalism</em>, <em>0</em>(0), 1–24. <a href="https://doi.org/10.1080/21670811.2023.2211626">https://doi.org/10.1080/21670811.2023.2211626</a></li><li>Nielsen, R. K., &amp; Ganter, S. A. (2022). <em>The Power of Platforms: Shaping Media and Society</em>. Oxford University Press.</li><li>Simon, F. M. (2024). <em>AI Search and Implications for Publishers </em><a href="https://felixsimon.medium.com/ai-search-implications-for-publishers-and-the-public-sphere-db931b6c8274">https://felixsimon.medium.com/ai-search-implications-for-publishers-and-the-public-sphere-db931b6c8274</a></li></ul><h4>AI, journalism and the public arena</h4><ul><li>Simon, F. M. (2024). <em>Artificial Intelligence in the News. How AI Retools, Rationalizes, and Reshapes Journalism and the Public Arena</em> (p. 46). Tow Center for Digital Journalism, Columbia University. <a href="https://www.cjr.org/tow_center_reports/artificial-intelligence-in-the-news.php">https://www.cjr.org/tow_center_reports/artificial-intelligence-in-the-news.php</a></li><li>Jungherr, A., &amp; Schroeder, R. (2023). Artificial intelligence and the public arena. <em>Communication Theory</em>, <em>33</em>(2–3), 164–173. <a href="https://doi.org/10.1093/ct/qtad006">https://doi.org/10.1093/ct/qtad006</a></li></ul><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=691053180d68" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[My Favourite Books of 2023]]></title>
            <link>https://felixsimon.medium.com/my-favourite-books-of-2023-1625a5ce039a?source=rss-41832dfb135a------2</link>
            <guid isPermaLink="false">https://medium.com/p/1625a5ce039a</guid>
            <category><![CDATA[books]]></category>
            <category><![CDATA[journalism]]></category>
            <category><![CDATA[technology]]></category>
            <category><![CDATA[reading]]></category>
            <category><![CDATA[ai]]></category>
            <dc:creator><![CDATA[Felix M. Simon]]></dc:creator>
            <pubDate>Fri, 15 Dec 2023 14:58:55 GMT</pubDate>
            <atom:updated>2023-12-15T14:58:55.129Z</atom:updated>
            <content:encoded><![CDATA[<figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*YggDrIXlXrxHPY6L45C5bg.png" /></figure><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*btE1SAsdunB4B4Bn3yeEcA.png" /><figcaption>A DALL:E variation on “The Hunters in the Snow” by Pieter Breugel the Elder</figcaption></figure><p>Every year, I list some of my favourite new Christmas albums as well as a couple of books I particularly enjoyed over the last year (see also the <a href="https://felixsimon.medium.com/o-come-all-ye-curious-6ff9c3d3c44b">2019</a>, <a href="https://felixsimon.medium.com/my-favourite-books-of-2020-d3f04f1c4665">2020</a>, <a href="https://medium.com/@felixsimon/my-favourite-books-of-2021-ef13ed06eb29">2021</a>, <a href="https://felixsimon.medium.com/my-favourite-books-of-2022-c6f24f588039">2022</a> editions). This year, it’s an unusually short list: not because I did not get round to reading, more because I am a bit too exhausted at this point to spend many of my remaining brain cells before Christmas to write about what I read. So only five recommendations this time, I’m afraid. A very happy Christmas to you all from Oxford and happy reading &amp; listening 🎄</p><h4>Books</h4><p><strong>Trading at the Speed of Light: How Ultrafast Algorithms are Transforming Financial Markets<br></strong><em>Donald MacKenzie, 2021, Princeton University Press</em></p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*kMI4MNK6_kTh39qiwy15cA.jpeg" /></figure><p>While not a book about AI — the area in which I’ve carved out my niche since before it become cool in 2019 — Donald MacKenzie’s book on high-frequency trading is the kind of book that should be written about AI, especially as it pertains to the news.</p><p>MacKenzie is a sociologist and it is fascinating to see how he is undaunted by the apparent complexity of the financial and algorithmic systems he seeks to explore, understand, and explain as part of his research (somewhere at the beginning of the book there’s a wonderfully wry remark about how all these things usually aren’t as complicated as they seem and that a lot of the talk around it is merely to keep people out).</p><blockquote>For those unfamiliar with what high-frequency trading is: think super-fast, computer-driven stock trading where algorithms make lightning-quick decisions to buy or sell for tiny profits in the blink of an eye. The faster you can process data and execute trades, the more likely that you can outperform your competitors. And there is an upper limit to that: the speed of light. Nothing can travel faster than light in a vacuum (approximately 300,000 kilometres per second) although you can get very, very close. One of MacKenzie’s examples is the data transmission time from the Chicago Mercantile Exchange’s datacenter to Nasdaq’s datacenter in New Jersey (in milliseconds (ms), one thousandth of a second) along the geodesic. With glass fiber (at a refractive index of 1.47) this takes about 6.65 ms. The fastest microwave links in 2016 needed about 3.98 ms. The Einsteinian limit for the same distance is about 3.94 ms.</blockquote><p>He’s not one to get sucked into bull-shitty talk of “technology revolutionising XYZ” but instead carefully lays out how the interplay of technological advances, financial incentives, human ingenuity, and cold, hard politics and power have re-shaped the infrastructure on which modern trading happens; and how trading itself has (and has not) changed as a result.</p><p>Come for the sociological framing…</p><p><em>“It would, for example, be a mistake to think of finance’s large technical systems as either simply the outcome of human intentions or as fully under human control. Such systems can manifest behavior that is neither anticipated nor welcomed by the human beings who design and construct them.</em>”</p><p>…stay for the crazy stories from the field, including this one about honey, bird-droppings, and designing a casing for a laser transmitter needed for a weather-resistant network of integrated laser/millimeter-wave links among the New Jersey share-trading network:</p><p><em>“High tech as that is, an utterly mundane material phenomenon threatened to cause a substantial difficulty: with the equipment needing to be the open air, on towers or on top of tall buildings, it was inevitable the bird droppings would fall on the glass of the laser units, potentially blocking signals. As Persico told me:</em></p><p><em>There were literally seven figures [a sum in excess of $1,000,000] spent on creating a [coating] that was resistant to bird droppings, and the way that [the bird droppings were] simulated [in testing coatings] is that honey has the same viscosity [and it was poured] on the units to develop a [coating] where that would slough off.”</em></p><p><strong>In Broad Daylight: Movies and Spectators After the Cinema<br></strong><em>Gabriele Pedulla, 2012, Verso</em></p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*xAaLo2vAN0qIDxWXsSX_hw.jpeg" /></figure><p>First published in 2012, “In Broad Daylight” is a great, and in my view time-tested exploration of the ever-evolving relationship between cinema and its audience. Pedulla skillfully navigates the complex intersections of technology, culture, and spectatorship, and chronicles how the “cinematic experience” has transcended the traditional confines of the movie theater— moving from a communal setting to personalised and fragmented experiences enabled by digital media.</p><p>The book isn’t particularly long but packed with punchy chapters that theorise the changing nature of film and cinema. What I found delightful about it was the absence of any form of nostalgia in Pedulla’s analysis. He clearly loves cinema and movie theatres but this doesn’t cloud his view when it comes to examining how technology and audience habits change the movies. For instance:</p><p><em>“Things changed as soon as technological innovations began to offer a wide range of possibilities and moving images were liberated from the picture house’s constraints. Precisely because it has been openly challenged, the movie theatre suddenly appears right before our eyes. Only now that the frame has shed its false naturalness, thanks to the competition of new media, are we able to see it as an artificial construction that was perfected over the course of decades.”</em></p><p>And:</p><p><em>“Following the auditorium’s decline, the style of films will change as well, and with it possibly the type of pleasure and aesthetic experience sought from moving images. Divested of the big screen, cinema of the future will inevitably be different from what we have had until now. As will its spectators.”</em></p><p>Words to remember, not least for those who currently theorise what AI “means” for [enter your favourite subject here].</p><p><strong>The Man From The Future: The Visionary Life of John von Neumann<br></strong><em>Ananyo Bhattacharya, 2021, Penguin</em></p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*74nLGRZBhzZ_iqEvQAzDmQ.jpeg" /></figure><p>I am not a fan of hagiography (which this arguably is, at least to a degree), and the book can be a bit dry at times but what Bhattacharya has achieved here is a fine and rich portrait of someone who was — by all accounts—nothing short of a genius, a true Renaissance man. And yet, many will have never heard of John von Neuman (even this year’s <em>Oppenheimer </em>somehow managed to ignore him).</p><p>A polymath of unparalleled depth, the Hungarian-American’s virtuosity extended far beyond the realms of mathematics, with his fingers dancing upon the keys of quantum mechanics, game theory, economics and computer science (and the book does a good job at taking us through his inventions and putting the things he contributed to in context). For example, have you ever wondered where the term “zero-sum” comes from?</p><p><em>Von Neumann could not get any further with multi-player games, so he switched to thinking about a situation with just two opponents whose individual payouts sum to zero. ‘It is not enough to succeed. Others must fail, Iris Murdoch once wrote. Von Neumann coined the term ‘zero-sum’ to describe such games of total conflict, in which one person’s loss is the other’s gain. One indication of the influence of game theory is that ‘zero-sum’ has now passed into the vernacular.</em></p><p>Even Wikipedia struggles to adequately convey just how influential his man was…</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*IcBG0P9d86obkoNKvmUlvg.png" /></figure><p>This, of course, is the other side of the coin: von Neuman played a key role in the invention of nuclear weapons, was part of the target selection committee, and spent much of his later life consulting for the US defence apparatus. With light, there’s also always shadow.</p><p>I’ll leave you with this quote from the last page of the book, which has a certain Rortyesque pragmatism to it:</p><p><em>‘For progress there is no cure. Any attempt to find automatically safe channels for the present explosive variety of progress must lead to frustration. The only safety possible is relative, and it lies in an intelligent exercise of day-to-day judgment.’</em></p><p><strong>AI &amp; Society: Spring 2022 Issue of Daedalus<br></strong><em>Edited by James Manyika, 2022, Journal of the American Academy of Arts &amp; Sciences</em></p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*Gpb7D6iDcfztTqax8X2atA.jpeg" /></figure><p>I do not have much to say on this book, other than that I have been recommending it to anyone who asked me about a recommendation for what to read on AI over the past year. The reasons are threefold:</p><p>a) It is available open-access <a href="https://www.amacad.org/sites/default/files/daedalus/downloads/Daedalus_Sp22_AI-%26-Society_0.pdf">here</a>.</p><p>b) It spans an impressive set of thinkers and views on the subject.</p><p>c) It has aged surprisingly well.</p><p>Tolle lege, tolle lege…</p><p><strong>Mantel Pieces: Royal Bodes and Other Writing from the London Review of Books<br></strong><em>Hilary Mantel, 2020, 4th Estate</em></p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*_VSEakO-g-Q43zMTEOiF6g.jpeg" /></figure><p>A compilation of brief essays, ‘Mantel Pieces’ is an ideal introduction if you’re not familiar with Hilary Mantel’s work (which everyone should be as far as I am concerned). Her lecture on <a href="https://www.lrb.co.uk/the-paper/v35/n04/hilary-mantel/royal-bodies">“Royal Bodies”</a> contained herein is legendary, of course, (and much misread as a personal attack on what was then Catherine, Duchess of Cambridge) although perhaps the best essay, “Meeting the Devil” focuses on Mantel’s recovery from surgery and offers a sharp observation of what it <em>means</em> to be ill:</p><p><em>Illness strips you back to an authentic self, but not one you need to meet. Too much is claimed for authenticity. Painfully we learn to live in the world, and to be false. Then all our defences are knocked down in one sweep. In sickness we can’t avoid knowing about our body and what it does, its animal aspect, its demands. We see things that never should be seen; our inside is outside, the body’s sewer pipes and vaults exposed to view, as if in a woodcut of our own martyrdom.</em></p><p>(For a great review that captures the essence of Mantel’s literary journalism, <a href="https://stingingfly.org/review/mantel-pieces/">look here</a>).</p><p>Setting aside the content and the literary qualities of her work, Mantel’s writing also serves as a poignant reminder of elements frequently overlooked in the humdrum of academic life: the significance of individuals’ intricate internal worlds and experiences, the importance of fearlessness and rigor, and, above all, the quality of our prose itself — with writing being more than just a vessel for information, even though we often treat it as such.</p><h4>Other</h4><p>Not a book and also not music: Instead, an essay by film scholar David Bordwell on why, yes, “Die Hard” is a Christmas Movie (among other things): <a href="https://www.davidbordwell.net/blog/2023/12/04/die-hard-revived-an-entry-revisited/">https://www.davidbordwell.net/blog/2023/12/04/die-hard-revived-an-entry-revisited/</a></p><h4>Music</h4><p>In terms of music, I did not find a single new Christmas album that I liked (so if you have any recommendations, send them my way). But I loved <a href="https://open.spotify.com/album/70DXuZk6U1852nP2qdh1lU?si=NPnEN8VVTF6ACHXwjGmu2Q"><em>Londres </em>by Oxford-based Costa Rican artist Patiño</a> — which is an absolute gem (and very much not a Christmas album).</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*af6IYOjEDarHGfI7kUw8Uw.png" /></figure><p>Here’s a taster (with the sweltering summer vibes beating the cold, grey winter in an instant…):</p><iframe src="https://cdn.embedly.com/widgets/media.html?src=https%3A%2F%2Fwww.youtube.com%2Fembed%2Fp8F37v-5cYU%3Ffeature%3Doembed&amp;display_name=YouTube&amp;url=https%3A%2F%2Fwww.youtube.com%2Fwatch%3Fv%3Dp8F37v-5cYU&amp;image=https%3A%2F%2Fi.ytimg.com%2Fvi%2Fp8F37v-5cYU%2Fhqdefault.jpg&amp;key=a19fcc184b9711e1b4764040d3dc5c07&amp;type=text%2Fhtml&amp;schema=youtube" width="854" height="480" frameborder="0" scrolling="no"><a href="https://medium.com/media/0ff13af3d7a475767e1ba71c10a27005/href">https://medium.com/media/0ff13af3d7a475767e1ba71c10a27005/href</a></iframe><p><a href="https://www.oii.ox.ac.uk/people/felix-simon/"><em>Felix Simon</em></a><em> is a Dieter Schwarz Scholar at the Oxford Internet Institute (OII) and a </em><a href="https://towcenter.columbia.edu/content/felix-m-simon"><em>Knight News Innovation Fellow</em></a><em> at Columbia University’s Tow Center for Digital Journalism where he is researching AI in journalism and the news industry. He works as a research assistant at the Reuters Institute for the Study of Journalism. His tweets can be found under </em><a href="https://twitter.com/_FelixSimon_"><em>@_Felix Simon_</em></a><em>.</em></p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=1625a5ce039a" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[“No Technology Is Going To Take Over All Of Society”]]></title>
            <link>https://felixsimon.medium.com/no-technology-is-going-to-take-over-all-of-society-c70efd017b9d?source=rss-41832dfb135a------2</link>
            <guid isPermaLink="false">https://medium.com/p/c70efd017b9d</guid>
            <category><![CDATA[democracy]]></category>
            <category><![CDATA[ai]]></category>
            <category><![CDATA[journalism]]></category>
            <category><![CDATA[oxford]]></category>
            <category><![CDATA[technology]]></category>
            <dc:creator><![CDATA[Felix M. Simon]]></dc:creator>
            <pubDate>Sat, 17 Jun 2023 10:49:30 GMT</pubDate>
            <atom:updated>2023-06-17T10:49:30.014Z</atom:updated>
            <content:encoded><![CDATA[<h4>Exploring LLMs, Generative AI, and their Societal Implications</h4><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*TxOp1A1EGxIXe0CN8P7MtA.png" /><figcaption>Copyright: DALL:E</figcaption></figure><p>How do Large Language Models function and what are the key components that make them work? How and where are they used, including in the media? Can they contribute to the promotion of democratic values?</p><p>These were some of the key questions we explored at the recent symposium “Automating Democracy: Generative AI, Journalism, and the Future of Democracy”, held on June 16th, 2023, at Balliol College, University of Oxford. Supported by the Balliol Interdisciplinary Institute, the Institute for Ethics in AI, and the Oxford Internet Institute, participants delved into the technical intricacies and practical applications of these emerging technologies, with the aim to explore how they may (or may not) affect our media landscape and democratic culture.</p><p>Organised jointly by <a href="https://www.oxford-aiethics.ox.ac.uk/linda-eggert">Dr Linda Eggert</a>, an Early Career Fellow in Philosophy, and <a href="https://www.oii.ox.ac.uk/people/profiles/felix-simon/">Felix M. Simon</a>, a communication researcher and DPhil student at the Oxford Internet Institute, the symposium provided a platform for scholars, experts, practitioners and student to engage in robust and open debates about LLMs, Generative AI, and their implications for our world.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*CkvnklPy7lrDOuC4kASBVA@2x.jpeg" /></figure><figure><img alt="" src="https://cdn-images-1.medium.com/max/512/1*qw6YVMRnYFdjbS6sjA02aw@2x.jpeg" /><figcaption>Felix Simon and Linda Eggert introducing the first panel</figcaption></figure><h4><strong>No Ghost in the Machine</strong></h4><p>The first panel focused on how Large Language Models (do not) work. The impact of ChatGPT has been so large primarily due to its user-friendly nature. “Current AI is often just re-branding. The progress of this technology has been more gradual than commonly assumed,” said Hal Hodson, a technology journalist at <em>The Economist</em>. Much of this remained obscured from public view, not because it was impossible to learn more about it, but because the broader public did not pay a lot of attention.</p><p>Hodson and Hannah Kirk, an AI researcher and DPhil student at the Oxford Internet Institute, concurred that Large Language Models (LLMs) are fundamentally intricate statistical machines, drawing on extensive datasets to generate output. Hodson, who has extensively covered the technology, argued that the vast majority of experts in the field view LLMs as machines, devoid of human-like attributes — contrary to what some recent news coverage might suggest. He also cautioned against overblown assumption about their impact: “No technology is going to take over all of society.”</p><p>Kirk delved into a less explored aspect — the refinement of these systems beyond initial their training. She emphasised the significance of reinforcement learning through human feedback (RLHF), a technique widely adopted by industry labs. However, Kirk expressed concern about the limited representation of these feedback providers and how their values profoundly shape AI systems. “We need to ask: Who are these people and how do their values end up shaping these systems?”</p><p>During the discussion, another important point emerged: the question of who reaps the rewards from these systems. As much of the pre-training relies on web-scraped data without compensating its creators, issues of copyright and remuneration claims loom large. Hodson anticipates an ongoing battle, yet believes that a concord between copyright holders and technologists will eventually be reached.</p><p>One common issue, recognised by both Kirk and Hodson, is the tendency to anthropomorphise these systems. “You cannot avoid anthropomorphisation but you should try,” said Kirk. They both strive to avoid this inclination in their work. Kirk, however, also noted that some companies have little incentive to discourage it, as it enhances the perception of their systems’ power beyond reality.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*DD_uD-hkFufWOXrjSZoL-w.jpeg" /><figcaption>Linda Eggert, Hannah Kirk, Hal Hodson</figcaption></figure><h4>AI is Not A Quick Fix For the Media</h4><p>One area where LLMs are increasingly used is the media and the news. “We began training individuals to use ChatGPT in 2018,” shared Laura Ellis from the BBC, highlighting the broadcaster’s early engagement with the technology. Since then, the organisation has experienced the typical upheavals that accompany any new technological advancement — enthusiasm mixed with concerns about job displacement and worries about the transformation of the BBC’s work. These days, the broadcaster adopts a cautious approach, actively considering safety, potential harms, and copyright issues, among a range of issues according to Ellis.</p><p>Gary Rogers, a media consultant and founder of news agency RADAR, emphasised the longstanding presence of AI in the news industry. “It is not a novel concept. Many people were simply unaware because it was primarily used downstream, for instance, in recommendations. This has changed with the advent of ChatGPT.” RADAR extensively uses natural language processing (NLP) and natural language generation (NLG) techniques, producing approximately 150,000 stories annually based on local and national data sources, covering topics such as local elections.</p><p>While a growing number of media organisations admit to exploring and using LLMs in their work, for example to write summaries or create new products, few have policies in place which guide their use. Rogers urged publishers to rectify this: “Organisations need to put policies in place so people can work with this safely, because they are already using this.”</p><p>Gemma Newlands, a sociologist and lecturer at the Oxford Internet Institute, provided a broader perspective on the adoption of new technologies within organisations and the accompanying challenges. She underscored the inherent tensions between the individual and organisational motivations for the use of any technology, emphasising that questions of power and agency cannot be avoided.</p><p>“Using ChatGPT can be individually benefitting for a journalist, but detrimental for the organisation if it introduces, for example, errors in journalistic copy,” Newlands argued. Conversely, organisations may devalue journalists’ work by relying excessively on such technologies, resulting in a situation beneficial to the organisation but detrimental to the individual. Newlands also contemplated how the adoption of AI could shape the social perception of journalism and how people value and trust it. And: “Finding a way to use AI is easy. Finding a way that makes sense for your business model is really hard.”</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*Ree3gs76VRmwC7HFAoMkiA.jpeg" /><figcaption>Linda Eggert, Gemma Newlands, Gary Rogers, Laura Ellis</figcaption></figure><h4>AI and Democracy or: Who Gets To Decide The Direction</h4><p>The final panel considered the impact of AI on democracy and the governance of the technology. Hélène Landemore of Yale University stressed the need to democratically build and regulate AI systems. “We have all the models to build a global assembly to think about regulating something like AI,” she argued, pointing to the success of citizen assemblies in thinking about ways to address issues such as climate change or assisted dying. “We need a global demos that goes beyond nation states to shape the development of this technology.”</p><p>John Tasioulas of the Institute for Ethics in AI warned that AI cannot be left to progress without adequate regulation. “The assumption that a technology will automatically be democratic in nature is a pipe-dream.” He sees AI as part of a larger tug-of-war between technocratic and more populist forces, with a compromise of some form needed. He was also critical of the goals towards which AI systems are currently optimised and developed. “The focus is often only on economic growth, but that is often not in the common interest.” AI should be developed for the common good, not the interests of a set of companies or technocratic elites.</p><p>Polly Curtis of the London-based think tank Demos shared some of these worries. “AI at the moment is an investment arms race powered by greed,” she argued. For Curtis, there is a chance to use AI to repair the relationship between citizens and the state in countries such as the UK, by making information and services more accessible and fostering a better connection between the state and its citizens. “But it could also go the other way, by baking in the biases and discrimination we already have. The lesson from the last 20 years of technological development is this: The worst is not going to happen but change is inevitable and will not necessarily positive.”</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=c70efd017b9d" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[AI in News Organizations: Exploring Applications, Challenges, and Future Implications]]></title>
            <link>https://felixsimon.medium.com/ai-in-news-organizations-exploring-applications-challenges-and-future-implications-fc667c695ab1?source=rss-41832dfb135a------2</link>
            <guid isPermaLink="false">https://medium.com/p/fc667c695ab1</guid>
            <category><![CDATA[oxford]]></category>
            <category><![CDATA[journalism]]></category>
            <category><![CDATA[ai]]></category>
            <category><![CDATA[news]]></category>
            <category><![CDATA[technology]]></category>
            <dc:creator><![CDATA[Felix M. Simon]]></dc:creator>
            <pubDate>Fri, 26 May 2023 15:52:05 GMT</pubDate>
            <atom:updated>2023-05-26T15:52:05.744Z</atom:updated>
            <content:encoded><![CDATA[<h3><strong>AI in News Organisations: Exploring Applications, Challenges, and Future Implications</strong></h3><h4><em>A symposium at Balliol college gathered insights from experts on AI’s role in news organisations and the implications of the technology for our information ecosystem</em></h4><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*lQ4YQZ77EQUYfE4DfEhCkA.png" /></figure><p>On the 25th of May 2023, a half-day symposium on the use of AI in news organisations took place within the serene setting of Balliol College, seeking to foster discussions between industry experts, academics, and students.</p><p>Organized and moderated by Felix M. Simon of the Oxford Internet Institute, the eventfeatured three panel discussions on ‘AI in the Workplace’, ‘AI, News, and the Role of the Technology Sector’, and ‘AI and the Implications for Public Discourse’ The event was funded by the TORCH Minderoo Challenge Fund in AI Governance and received generous administrative support from the Oxford Internet Institute and Balliol College.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*5vwKt1Qp0Ly-iRXSG9RgFg.jpeg" /></figure><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*rOuT-PdADxWPNLxb537yBg.jpeg" /></figure><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*1OVCeGsU5objyA9b_P4f2Q.jpeg" /></figure><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*BUTeTPk87PEcZDm1VH5YpA.jpeg" /><figcaption>Image credit: Felix Simon</figcaption></figure><p>Participants included leading experts on AI and the news, including Shreya Vaidyanathan, Product Manager at <em>Bloomberg LP</em>; Jane Barrett, Global Editor for Media News Strategy at <em>Reuters</em>; Siddharth Venkataramakrishnan, Banking and Fintech Correspondent at the <em>Financial Times</em>; Melissa Heikkilä, Senior Reporter for AI at <em>MIT Technology Review</em>; Nic Newman, Senior Research Associate at the <em>Reuters Institute for the Study of Journalism</em>; David Caswell, Executive Product Manager at the <em>BBC</em>; and Tom Standage, Deputy Editor at <em>The Economist</em>. The diverse audience comprised journalists, policy experts, technology specialists, and students from the University of Oxford and beyond, contributing to a rich and multifaceted exchange of ideas.</p><p>As the event was held under Chatham House Rule, this brief summary provides a collective perspective, with a more extensive report to follow.</p><h3><strong>Looking Beyond Generative AI in the News</strong></h3><p>The speakers highlighted how AI in news organizations extends beyond generative AI and encompasses a wide range of applications. For example, AI is increasingly used for high-speed data extraction and analysis, in investigative journalism, as well as for optimizing existing news products and creating new ones. A significant portion of AI implementation is focused on non-audience facing tasks, such as organizing archives and content (although AI also looms large in the recommendation of content and the optimization of paywalls).</p><p>One connecting them was that claims regarding efficiency and productivity gains through AI should be approached with caution, as the extent of these gains varies depending on the specific task and context. While efficiency gains may be more straightforward in e.g. coding, they are not as easily achieved in tasks such as writing or reporting. In fact, the use of AI can sometimes lead to additional work for journalists and news workers. Some speakers also highlighted that the fear of job replacements by AI is not a simple binary scenario. Jobs in journalism consist of a combination of tasks, only some of which can be automated. This realization should provide some reassurance for journalists, as it indicates that their roles are not entirely replaceable by AI.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*crlLBgr0okDWWGgBrBnLfw.jpeg" /></figure><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*givFOrOt7Pvf9bLADZkQbA.jpeg" /><figcaption>Left image: Siddharth Venkataramakrishnan (FT), Tom Standage (The Economist), Shreya Vaidyanathan (Bloomberg), Jane Barrett (Reuters). Image credit: Felix Simon</figcaption></figure><h3><strong>No Easy Way Out of Big Tech’s Dominance Over AI</strong></h3><p>During the discussion on news organisations’ dependency on the technology sector for AI, participants engaged in a lively exchange of ideas. A key point raised was the dominance of Silicon Valley-based technology companies in AI research and development. These companies were likened to ‘landlords’ by one panellist as they exercise control over crucial resources such as data, computing infrastructure, and substantial financial and human capital, thus both shaping and controlling large swathes of the AI ecosystem. In contrast, smaller start-ups like OpenAI were described as ‘tenants’ heavily reliant on these larger firms for their AI development.</p><p>Another participant emphasised how big tech companies not only drive the adoption of AI but also already exert control over access to news audiences and traffic to news sites — something that potentially stands to be extended as AI is rolled more broadly. ‘Users have voted with their feet already’, gravitating towards the convenience and user-friendly experiences offered by these platform companies, solidifying their position in the news ecosystem. A shared concern in this context was how this broader development would not only affect news organisations’ business models but also the information ecosystem at large.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*F1pw9G8l34FgNYmQ7f3glg.jpeg" /></figure><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*7rD89qouIdGE_wRgmsHThA.jpeg" /><figcaption>Left image: Felix Simon (OII), Nic Newman (Reuters Institute), <em>Melissa Heikkilä </em>(MIT Tech Review), David Caswell (BBC). Image credit: Hicham Yezza and Felix Simon</figcaption></figure><p>The panel acknowledged that the development and implementation of AI would further consolidate the power of these companies unless proactive measures were taken. However, there were diverging opinions on the appropriate course of action and the responsible parties involved. Suggestions included strengthening competition and data protection laws, although it was acknowledged that regulators often struggle to keep pace. Additionally, panellists stressed the importance of news organisations confronting these challenges directly and proactively.</p><h3>‘<strong>Listen to the Generalists, Not Just the Computer Scientists’</strong></h3><p>The closing discussion involving Tom Standage of the <em>Economist</em>, Felix Simon, and the other participants and attendees synthesised the key themes explored throughout the day, while also offering broader insights into the future of AI, news, and information ecosystems.</p><p>One topic of discussion was the potential democratisation of AI through open source models, which can provide powerful functionalities at a low cost. It was also noted that we may not have reached the full potential of Large Language Models, suggesting that further advancements are still on the horizon. Another perspective presented was that in an increasingly noisy information environment flooded with cheaply produced content of questionable quality, the value of publishers as trusted brands may rise — an optimistic view of AI’s impact.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*yMz9Q2D8vx53V2ZXk7QqPw.jpeg" /><figcaption>Felix Simon and Tom Standage. Image credit: Hicham Yezza</figcaption></figure><p>However, the central point of the closing remarks emphasised the difficulty of making accurate forecasts about the future of AI and its impact and cautioned against overstating the likelihood of major risks associated with AI. The audience was encouraged to consider insights from generalists, not solely relying on computer scientists who develop AI.</p><p><em>For questions, please feel free to reach out to Felix at: </em><a href="/service/mailto:felix.simon@oii.ox.ac.uk"><em>felix.simon@oii.ox.ac.uk</em></a></p><p>About the Author:</p><p>Felix M. Simon is a journalist, communication researcher, and doctoral student at the <a href="https://www.oii.ox.ac.uk/people/profiles/felix-simon/">Oxford Internet Institute (OII)</a>, a <a href="https://towcenter.columbia.edu/content/felix-m-simon">Knight News Innovation Fellow</a> at Columbia University’s Tow Center for Digital Journalism, and an <a href="https://citap.unc.edu/announcing-the-2022-23-citap-affiliate-community/">affiliate at the Center for Information, Technology, and Public Life</a> (CITAP) at the University of North Carolina at Chapel Hill. He also works as a <a href="https://reutersinstitute.politics.ox.ac.uk/people/felix-simon">research assistant at the Reuters Institute for the Study of Journalism (RISJ) .</a></p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=fc667c695ab1" width="1" height="1" alt="">]]></content:encoded>
        </item>
    </channel>
</rss>