<?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 Agnes Stenbom on Medium]]></title>
        <description><![CDATA[Stories by Agnes Stenbom on Medium]]></description>
        <link>https://medium.com/@agnes.stenbom?source=rss-a5cd57d97916------2</link>
        <image>
            <url>https://cdn-images-1.medium.com/fit/c/150/150/2*ZbLpdC26UIkzk03lagfV9Q.jpeg</url>
            <title>Stories by Agnes Stenbom on Medium</title>
            <link>https://medium.com/@agnes.stenbom?source=rss-a5cd57d97916------2</link>
        </image>
        <generator>Medium</generator>
        <lastBuildDate>Fri, 20 Nov 2026 14:52:26 GMT</lastBuildDate>
        <atom:link href="https://medium.com/@agnes.stenbom/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[Can generative AI help us promote diversity in journalism?]]></title>
            <link>https://medium.com/@agnes.stenbom/can-generative-ai-help-us-promote-diversity-in-journalism-c9311bf72ddd?source=rss-a5cd57d97916------2</link>
            <guid isPermaLink="false">https://medium.com/p/c9311bf72ddd</guid>
            <category><![CDATA[journalism]]></category>
            <category><![CDATA[midjourney]]></category>
            <category><![CDATA[dalle-2]]></category>
            <category><![CDATA[stable-diffusion]]></category>
            <category><![CDATA[generative-ai]]></category>
            <dc:creator><![CDATA[Agnes Stenbom]]></dc:creator>
            <pubDate>Thu, 08 Dec 2022 10:04:18 GMT</pubDate>
            <atom:updated>2022-12-08T10:11:29.674Z</atom:updated>
            <content:encoded><![CDATA[<h4>By Agnes Stenbom &amp; Delfina Arambillet</h4><p><em>In the wake of groundbreaking developments within visual generative artificial intelligence (AI), a new debate about bias and model harms has emerged. What might this all mean for journalism? This blogpost summarises discussions held by team members behind the AIJO Project — a global group dedicated to a more diverse and inclusive future for the field.</em></p><h3>Background: The AIJO project</h3><p><a href="https://www.aijoproject.com/">The AIJO project </a>was a collaborative project run in 2020 as part of the JournalismAI Collab of LSE Polis. Representatives from eight major news organisations across the world came together to explore a big challenge: How might we leverage the power of AI to understand, identify and mitigate newsroom biases? We explored the topic by, among other things, <a href="https://www.aijoproject.com/identify">doing an AI-driven analysis of the gender balance in our reporting. </a>We focused primarily on images, as it was the language of expression we all had in common.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*ylyp6cE9ANiI1FpOOmFXXw.png" /><figcaption>The AIJO Project first launched in 2020</figcaption></figure><p>AIJO is an abbreviation for <strong>AI + Jo</strong>urnalism. Coincidentally, as we learned from our team members from Nikkei, Aijō / 愛情 also means love in Japanese.</p><p>We believe that there is great potential in using AI tools to create more compassionate journalism that celebrates and manifests diversity, and our hope was for our 2020-work to contribute to such a future. Now in 2022, we see the need for a related industry discussion about a topic with huge potential future impact for journalism and the broader media field: generative AI.</p><h3>Why care about generative AI?</h3><p>Generative AI has been <a href="https://www.techopedia.com/definition/34633/generative-ai">defined</a> as a<em> “broad label that’s used to describe any type of artificial intelligence that uses unsupervised learning algorithms to create new digital images, video, audio, text or code.” </em>While the developments within all of these domains are exciting, we find image-based approaches especially interesting to discuss given the fact that they are already generating high-quality output which is being leveraged in digital media applications across the world. Unlike e.g. natural language processing, there are no language barriers limiting adoption to certain markets or regions.</p><p>The perhaps most popular visual generative AI models are <strong>DALL·E 2 </strong>(OpenAI),<strong> Stable Diffusion </strong>(Stability.ai), and <strong>Midjourney</strong>. These models can be used as stand alone services, or as part of other. For example, Microsoft is adding DALL·E 2 to <a href="https://info.microsoft.com/ww-landing-ai-interest-capture-execution-for-ignite-keynote.html?lcid=en-usus%2Ffree%2Fai%2Fsearch%2F&amp;utm_campaign=The%20Batch&amp;utm_medium=email&amp;_hsmi=232281683&amp;_hsenc=p2ANqtz-9gkkSJTFR8E0u99iWYB0VNudjHn59i83z83ZJSzoALg6wwwi7LHvkbdUx5nGcAv4KHuHOyn-rbpMpqKfzbzbdyScXvxPf95V4Ro0GajX4qGZe0RJQ&amp;utm_content=232281151&amp;utm_source=hs_email">Azure OpenAI</a> and Designer (an app that automates graphic design for e.g. social media), and Shutterstock will <a href="https://www.theverge.com/2022/10/25/23422359/shutterstock-ai-generated-art-openai-dall-e-partnership-contributors-fund-reimbursement?_hsmi=232281683&amp;_hsenc=p2ANqtz-8A3HQo6sWWgcSxK2gXPIOXh4n9CkwN02Kso-Nt3spS94TD4fSO3TmXkO0zHvVQlWhd9qNHKjpHA7AeCZqb0lhwdO3RCytphc8aZ4yUbKnHXzGSD8Y">allow</a> users to generate images using the same model.</p><h3>AI is inherently conservative</h3><p>To convert text to images, generative models are trained with thousands of images (often from open images from the internet) and their corresponding captions. These models are hence trained on historical data, and with that comes risks of perpetuating biases and stereotypes of the past. But what does that mean in practice?</p><p>Let’s use a few areas where journalism has well-known traditional biases to exemplify.</p><p>Imagine a journalist covering the stock market. Who do you picture?</p><p>Our bet is that they look something like this:</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*G7t-8_UNXZrdO9EYmXfQyQ.png" /><figcaption><em>Stable Diffusion, prompt: “A journalist covering the stock market”</em></figcaption></figure><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/0*zfdkb010vLOuyu1O" /><figcaption><em>DALL:E, prompt: “A journalist covering the stock market”</em></figcaption></figure><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*y_oKbia08keEF8Ltw8T7gw.png" /><figcaption><em>Midjourney, prompt: “A journalist covering the stock market”</em></figcaption></figure><p>Let’s try fashion! Who is a fashion journalist according to AI?</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*fuEHe2UaKJH3ztxakh5QBQ.png" /><figcaption><em>Stable Diffusion, prompt: “A journalist covering fashion”</em></figcaption></figure><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/0*jjeM59tJQevkugNA" /><figcaption><em>DALL:E, prompt: “A journalist covering fashion”</em></figcaption></figure><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*ezUPsVj7lAWp1cwZtp2TwA.png" /><figcaption><em>Midjourney, prompt: “A journalist covering fashion”</em></figcaption></figure><p>What if we remove the topic and just ask for “<strong>a journalist</strong>”? Seems like AI prefers such in formal attire.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*Xew9UbUcigbXbOnYzlw-BA.png" /><figcaption><em>Stable Diffusion, prompt: “A journalist”</em></figcaption></figure><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/0*ABIFhSuyA2kAeWup" /><figcaption><em>DALL:E, prompt: “A journalist”</em></figcaption></figure><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*RMWgV7SaEOCQeTkHvnQgSQ.png" /><figcaption><em>Midjourney, prompt: “A journalist”</em></figcaption></figure><p>These examples are illustrative and fairly low-stakes. But in reality, the consequences of machine systems with e.g. gender, racial, or age biases encoded into them can have serious consequences in regards to how and whom we depict. Representation matters.</p><h3>Known shortcomings</h3><p>We have known about bias in AI and machine learning for quite some time now, with seminal research being made by researchers like Dr. Joy Buolamwini, Dr. Margaret Mitchell, Dr. Timnit Gebru, and Inioluwa Deborah Raji. Many tech companies have also been developing areas devoted to Diversity, Equity &amp; Inclusion in order to bring these topics to the table.</p><p>In the domain of generative AI, initiatives like the <a href="https://huggingface.co/spaces/sasha/StableDiffusionBiasExplorer">Stable Diffusion Bias Explorer</a> has put the spotlight on how models can generate biased imagery.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*7p58ioqZL3SDPx5uWd9YOQ.png" /><figcaption>The <a href="https://huggingface.co/spaces/sasha/StableDiffusionBiasExplorer">Diffusion Bias Explorer</a></figcaption></figure><h4>Model documentations and policies</h4><p>The creators of the models are not blind to these issues either. The documentation of <strong>Stable Diffusion, DALL·E 2 </strong>and<strong> Midjourney </strong>all include concerns about the biases and harms the models may perpetuate, and they companies behind them have developed different ways of mitigating them. Let’s explore what they say.</p><p><a href="https://openai.com/dall-e-2/"><strong>DALL·E 2</strong></a><strong><br></strong><em>Developed by OpenAI, an AI research laboratory founded by Elon Musk and Sam Altman, Microsoft majority owner.</em></p><p>DALL·E 2 may well be the most popular generative model and you might have seen lots of its generated images on Twitter. According to their webpage, this 2.0 version of the DALL·E model has incorporated “<em>trust &amp; safety lessons we’ve learned while deploying DALL·E to 3 million artists and users worldwide, developers can ship with confidence knowing that built-in mitigations — like filters for hate symbols and gore — will handle the challenging aspects of moderation.” </em>In a <a href="https://openai.com/blog/dall-e-2-pre-training-mitigations/">blogpost</a> titled “<strong>Pre-Training Mitigations”</strong>, OpenAI write that <em>“in order to share the magic of </em><a href="https://openai.com/dall-e-2/"><em>DALL·E 2</em></a><em> with a broad audience, we needed to reduce the risks associated with powerful image generation models.”</em> These mitigations are said to include:</p><ul><li>Filtering Graphic and Explicit Training Data</li><li>Fixing Bias Introduced by Data Filters</li><li>Preventing Image Regurgitation</li></ul><p><a href="https://stability.ai/blog/stable-diffusion-announcement"><strong>Stable Diffusion</strong></a><strong><br></strong><em>Developed by stability.ai, the world’s first community-driven, open-source artificial intelligence company.</em></p><p>Stable Diffusion is a publicly released model (<a href="https://huggingface.co/spaces/stabilityai/stable-diffusion">test it here</a>) available under a<a href="https://huggingface.co/spaces/CompVis/stable-diffusion-license"> Creative ML OpenRAIL-M licens</a>; a permissive licence that allows for commercial and non-commercial usage. This licence is <strong>focused on ethical and legal use of the model </strong>and, as they mention in their webpage, must accompany any distribution of the model.</p><p>An AI-based<strong> Safety Classifier</strong> included by default in the software package, which understands concepts and other factors in generations to remove outputs that may not be desired by the model user.As they write, <em>“the parameters of this can be readily adjusted and we welcome input from the community on how to improve this. Image generation models are powerful, but still need to improve to understand how to represent what we want better.”</em></p><p><a href="https://www.midjourney.com/home/"><strong>Midjourney</strong></a><strong><br></strong><em>Developed by Midjourney, an independent research lab focused on design, human infrastructure, and AI</em></p><p>We have not identified any public documentation about model biases and/or training process. However, Midjourney has a content moderation process described at high level in <a href="https://midjourney.gitbook.io/docs/content-and-moderation-policy">their documentation</a>. The model is available through an open Discord-channel, where the policy for image generation is detailed in the #rules channel:</p><ul><li><em>“Do not create images or use text prompts that are</em><strong><em> inherently disrespectful, aggressive, or otherwise abusive. </em></strong><em>Violence or harassment of any kind will not be tolerated.”</em></li><li><em>“No adult content or gore, detached body parts of humans or animals, cannibalism, blood, violence. </em><strong><em>Please avoid making visually shocking or disturbing content. </em></strong><em>We will block some text inputs automatically.” </em>We find no documentation about what texts this would concern (i.e. which words are banned).</li></ul><p>They also point that they are working on a<em> “better moderating system. For now banning words has proven an effective way to prevent abuses. This list is under constant review. Some words sound safe but are bringing the model into unexpected corners in a consistent way.”</em></p><h3>Implications for journalism</h3><p>The intricacies of AI bias and encoded prejudices are poorly captured in slogans. But one basic conclusion is clear: AI won’t visualise a more diverse future for journalism unless we ask for it.</p><p>In our 2020-edition of the AIJO Project, we wrote the following:</p><blockquote>“The media cannot change who is in charge of countries, institutions and companies and we will continue to cover them regardless of race, gender or age. But we can decide how we report on the impact those leaders have on real people of all backgrounds and the way we represent diverse communities.”<em> </em><a href="https://www.aijoproject.com/post/understanding-bias-in-the-newsroom"><em>(AIJO, 2020)</em></a></blockquote><p>This of course still holds true, but we find it important to now also highlight how the ways we use emerging technologies (like those discussed in this blogpost) can have implications related to diversity and inclusion.</p><p>As shown in our short review above, the companies behind the models are concerned with the issues of bias and model harms. While we find that much of the responsibility to manage these issues falls on the model creators, we — the <strong><em>users</em></strong> of them — should also keep certain things in mind when exploring generative AI in and for journalism.</p><h4>Three suggestions for newsrooms</h4><p>This blogpost’s title asks a big question: Can generative AI help us promote diversity in journalism? We think it can, but it will require active measures from both creators and users of these models.</p><p>We have three suggestions for newsrooms:</p><ol><li><strong>Be mindful about the biases of the models. </strong>Avoid falling into stereotypes and take an active approach to prompting the models. If you e.g. want to generate a picture for an article about the stock market, consider adding ‘female’ to the prompt (the models will be conservative by default, remember?).</li><li><strong>Investigate the models</strong>. Leverage your newsroom’s investigation skills and seek further understanding of emerging models. What data were they trained on? How might they impact different communities of people? Report what you learn!</li><li><strong>Contribute to developing the models. </strong>If identifying shortcomings, evaluate wether you have the means to contribute to the field’s development by e.g. providing training data or development skills. Can you open source contributions, or in other ways pitch in to promote diversity and inclusion ideals in these developments?</li></ol><p>True to the original focus of the AIJO Project, this post has focused on visual generative AI models. However, it should be noted that generative <em>language</em> models are also making exponential progress at the moment. A key example is the newly released<a href="https://openai.com/blog/chatgpt/"> ChatGPT</a> model by OpenAI — a chatbot producing text at astonishing quality.</p><p><strong>We are excited about these developments and hope you will join our quest to explore the potential of generative AI in a manner that promotes diversity and inclusion.</strong> To give you some motivation to do so, we’d like to end with some wise words from ChatGPT:</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/867/0*NwPRbe-s1xm1c0L_" /><figcaption>ChatGPT on diversity and inclusion in journalism</figcaption></figure><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=c9311bf72ddd" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Hey Editor! It’s time to get involved in AI transparency]]></title>
            <link>https://medium.com/@agnes.stenbom/hey-editor-its-time-to-get-involved-in-ai-transparency-683640de5616?source=rss-a5cd57d97916------2</link>
            <guid isPermaLink="false">https://medium.com/p/683640de5616</guid>
            <category><![CDATA[artificial-intelligence]]></category>
            <category><![CDATA[journalism]]></category>
            <category><![CDATA[transparency]]></category>
            <dc:creator><![CDATA[Agnes Stenbom]]></dc:creator>
            <pubDate>Tue, 29 Nov 2022 12:51:46 GMT</pubDate>
            <atom:updated>2022-11-29T12:53:39.312Z</atom:updated>
            <content:encoded><![CDATA[<h4><em>By Agnes Stenbom, Kasper Lindskow and Olle Zachrison</em></h4><h4>Applied artificial intelligence has become an integral part of the day-to-day operations of many media companies, and new capabilities are emerging at stunning speed. But how informed are media consumers of what is going on? And who is accountable for the bigger picture? Looking back on discussions on AI transparency in the Nordic AI Journalism network, we see a need for increased leadership involvement.</h4><p><em>On October 4th 2022 the Norwegian public broadcaster NRK published a standard-looking article about energy prices. However, the credentials for the associated image were not of your average photographer. Instead, the depicted power lines and serene mountainous landscape were attributed to “Midjourney (AI)”.</em></p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*1pSd-1VIXW1s2jis0lbSjw.jpeg" /><figcaption>The initial illustration published by NRK</figcaption></figure><p><em>Did the byline make users understand and appreciate the synthetic nature of the content? The responsible editor at NRK didn’t seem convinced, and the image was swiftly taken down and replaced by a traditional stock photo. “There should be good reasons to use AI-generated illustrations as part of our news operations”, he explained in a subsequent </em><a href="https://journalisten.no/nrk-brukte-ai-illustrasjon-ved-en-feil-i-omtalen-av-strompriser/538674"><em>interview with Journalisten</em></a><em>.</em></p><p><em>We, three members of the Nordic AI Journalism network, find that this case highlights important strategic and ethical questions for media companies going forward.</em></p><p><a href="https://www.nordicaijournalism.com/">Nordic AI Journalism</a> is an industry network consisting of about 200 individuals working for more than 30 different news media organisations in the Nordics. The network seeks to contribute to responsible use and development of AI technologies in journalism by sharing learnings across organisational boundaries. The network was founded by Agnes Stenbom (Schibsted) and Olle Zachrison (Sveriges Radio) in 2020, and since then we have met virtually for bi-monthly sessions covering various cases related to AI and journalism — from synthetic voices to entity extraction models.</p><p>This autumn, we finally got to meet in person. With local hubs in Stockholm (hosted by Sveriges Radio), Copenhagen (Ekstra Bladet), and Oslo (Schibsted) we discussed <strong>AI transparency and why we need more user-accessible information about the use of AI in journalism.</strong></p><p>In this blogpost, we will highlight different types of transparency discussed, describe the consequences of opaque processes, and suggest a framework to chart use cases and risks going forward. Our key message: we see an urgent need for action, not least by media executives.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*w2cOe9dk8ReBlNwbglQwpw.png" /><figcaption>The Nordic AI Journalism network gathers around 200 news media professionals from across the Nordics</figcaption></figure><h3>Why AI transparency?</h3><p>First, we want to stress why we find the question of transparency related to the use of AI in news media to be such a pressing issue for journalism at this moment of time.</p><ol><li><strong>Transparency is intrinsically important for responsible AI practices</strong><br>Transparency in the use of AI towards the people who are affected by it is widely recognized as a central aspect of responsible AI practices, as it is the foundation for interacting with AI systems in informed ways as autonomous beings.</li><li><strong>Transparency to build trust</strong> <br>In journalism, we don’t yet have a ‘contract’ with media consumers about how and for what to use AI — neither as an industry or individual media companies. We find that openness is necessary to enable discussions among news publishers and stakeholders about the dos and don’ts of AI in order to establish more mature guidelines, policies and frameworks for the news domain. Specifically, we believe that increased transparency towards users would enable the discussion needed to facilitate expectation alignment — which is essential to facilitate trust.</li><li><strong>Transparency should be prioritised given the immature state of AI in news</strong><br>As highlighted by recent reports in and beyond the Nordics — see e.g. <a href="https://medierochdemokrati.lindholmen.se/nyheter/sa-tanker-mediechefer-i-norden-om-ai">Wiik</a> (2022) and <a href="https://drive.google.com/file/d/1utmAMCmd4rfJHrUfLLfSJ-clpFTjyef1/view">Beckett</a> (2019) — the news media industry is still very much at the start of our AI journey. A positive aspect of this is how it provides us with an opportunity to lay a solid foundation. We find that this foundation will be especially important going forward given that transparency in automated decision making and AI is mentioned in e.g. the GDPR, the Digital Services Act, and the AI Act.</li></ol><p>Contemporary, often open-sourced, AI models are producing staggering results and are already used at scale, both by the general public and journalists. Generative AI models are, as we speak, drafting quick versions of press releases and crafting synthetic video, audio and images, as mentioned above. <strong>These exciting developments add a sense of urgency to the importance of AI transparency.</strong></p><h3>Types and levels of transparency</h3><p>Given the great diversity in AI applications for news, from recommenders to internal tagging tools or generative content creation systems, we find that different types and levels of transparency are needed in different situations.</p><p>Informed by network discussions, we have distinguished between four different types and levels of transparency of relevance to news organisations:</p><ul><li><strong>Tell users that an AI system is being used (basic visibility) <br></strong>The perhaps most basic type of transparency involves informing users when any type of AI system is being used to select or modify content they are being served. This can be done in multiple ways, ranging from general AI policies that describe in general terms that AI systems are used by the news publisher, to specific information whenever AI is used in a specific element on a news website or app.</li><li><strong>Describe how the AI system makes decisions (technical visibility)<em><br></em></strong>Another type of transparency that builds on basic visibility is descriptions of how the algorithms that are being used work. This type of transparency may range from generic descriptions of the AI algorithms to detailed descriptions of each of the algorithms including what they optimise for, what input data is being used, what type of machine learning methods is used. These descriptions can be in technical terms or in layman’s terms.</li><li><strong>Explain individual decisions made by the AI system<br></strong>Yet another type of transparency involves explanations of the individual decisions that are made by AI systems, such as the reasons why a specific set of news articles are being served to a specific reader. This type of transparency ranges from post hoc explanations (where another AI algorithm guesses the reasons for a specific decision by examining what underlying features the decision correlate with, e.g. the readers gender or propensity to read e.g. sports stories) to intrinsic explanations where the algorithmic that makes the decision shows what internal factors caused the decision (this is typically more precise, but difficult for laymen to interpret).</li><li><strong>Enable users to directly affect individual decisions<br></strong>The final type of transparency we discussed involves enabling the reader to<strong> </strong>affect the AI algorithms they are being exposed to in order to build a deeper understanding of how the algorithm works. At one level this may involve simple opt in/opt out options where the users can choose between a version of a website or an element on a news website with or without AI systems. At another level, the user may be given “handles and levers” that influence the input to the AI algorithm allowing them to experiment with different settings and outputs.</li></ul><h3>A basic framework for charting use cases</h3><p>So how do you know which kind of transparency your specific use case calls for? In the network, we discussed two different factors that should impact such assessments.</p><p>First, we must consider the <strong>editorial risks</strong> associated with the use of AI. We find that it is good to make a distinction between legal risks (which cannot be negotiated) and ‘pure’ editorial risks. Sometimes they overlap, but there are certainly cases where a specific AI usage can create editorial risks that would not be a formal legal liability. A bad cluster of AI recommendations, or serious mistakes committed by machine translation can cause brand damage without any legal overstepping.</p><p>Secondly, we must consider the <strong>impact on the user experience</strong>. Is the user getting a different type of media experience because of the use of AI, or are we using the technologies in purely newsroom-facing processes, e.g. by making existing processes more efficient through automation?</p><p>In the network, we structured these two topics — editorial risks and impact on user experience — as axes to create a basic discussion framework:</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*2YhMWFYmbqIzwJ2iTfsbqg.png" /><figcaption>Our framework for facilitating discussions on AI transparency in news</figcaption></figure><p>Once a given team has agreed on where its use case places, we argue that they can have a more informed discussion about the need for informing users and/or other stakeholders.</p><h3>Risks come in different forms</h3><p>During the Nordic AI Journalism network meetups, we discussed a number of active use cases of AI in Nordic media from a transparency perspective. Should they have been communicated to users in a clearer way? If not, why not? If yes, how?</p><p>Most of the discussed use cases focused on recommender systems, but also included tagging of video content, audio transcriptions, and automated generation of news telegrams. Use cases included both ML-based systems and rule-based systems (described together below as “AI systems”). What they all had in common, though, was that none was communicated to users as an AI system.</p><p>When discussing our use cases in relation to the above framework, it quickly became clear that we assigned different levels of editorial risk to different types of use cases. In general, we were confident in our editorial process and decisions, and did not worry about AI e.g. recommending or translating content initially ‘vetted’ by news professionals. AI systems with higher degrees of automation, such as auto-generated news stories , were suggested to involve higher editorial risks, and with it, even higher demands for transparency.</p><h3>… but do users want it?</h3><p>It was argued by some in the network that users rarely express any interest in transparency and that they do not want direct control over AI systems (e.g. recommenders). To paraphrase one network member: <em>“They just express a desire for smooth, relevant experiences — they don’t care how it’s produced”.</em></p><p>Others in the network (including the authors of this blogpost) argue the opposite, with examples of users requesting clear information about who/what institution that is “behind” the AI, the data used to train it, and why AI is being applied in the first place.</p><p>Despite the internal differences in perceptions of users’ desires, most members have argued that more transparency <strong>is</strong> important and desirable — not least in order to increase internal focus on ethical practices that withstand the light of day.</p><h3>Implementation of AI across the journalistic process calls for transparency</h3><p>Our network discussions have highlighted the importance of recognizing how AI applications are part of a journalistic value chain where the number of use cases and the scale of their impact is growing. Sometimes one specific AI use case in itself can be assessed as low risk, but combined with other AI applications the picture rapidly becomes more complex and the risks of opaqueness are harder to calculate. <strong>No AI use case appears in a silo.</strong></p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*zBYzU1HwZRfLvk97Cf_A0Q.jpeg" /><figcaption><a href="https://unsplash.com/photos/s2TPKAD6fag">Art by Nidia Dias for DeepMind/Unsplash.</a></figcaption></figure><h3>Internal transparency as a basic requirement</h3><p>Keeping this in mind, we think that internal transparency and <strong>a shared understanding of how different AI use cases interplay</strong> is a crucial starting point for developing products that promote AI transparency in journalism. We need to understand the wider picture and inform users accordingly.</p><p>In modern media organisations, we argue that newsroom leaders — not least editors and/or publishers — must have an overview of where, when and how AI is impacting the production process and user experience. How else can the public hold them accountable?</p><p>It is easy to imagine a number of problems if media leaders do not become more aware about AI applied in their newsrooms. On the one hand, the risks of leadership overlooking the great potential of AI can hamper innovation in the media field. On the other, failing to recognize risks can have severe negative consequences for media companies and consumers alike.</p><p>Not being able to explain the general workings of systems applied in the own editorial process can be damaging both to the news brand and the personal credibility of the editor. While “human-in-the-loop” has become a popularized idea in AI, we believe we need to invest more specifically in making sure that we have “editor-in-the-loop”-systems as we continue to explore and apply AI in journalism.</p><h3><strong>Telling users that AI systems are at play: a needed first step</strong></h3><p>Finally, we would like to again stress how a basic level of transparency towards the readers is essential whenever AI systems directly impact the news experience without a human being in the loop, e.g. news recommender systems or content created (or translated) and published automatically. Such setups increase the need for information that allows readers to understand the news they are being served (and the errors AI systems might make) in context.</p><blockquote>The consumer has a right to be informed when AI is playing a crucial role in the media experience.</blockquote><p>In this digital information landscape, we believe that the consumer has a right to be informed when AI is playing a crucial role in the media experience. Exactly how this is done must be up to each company. However, our experience from the Nordic AI Journalism network is that a peer-to-peer dialogue about AI transparency, both within your company and the wider industry, is an excellent place to start.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=683640de5616" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Grand & Gradual: A call for AI Shepherds]]></title>
            <link>https://medium.com/@agnes.stenbom/grand-gradual-a-call-for-ai-shepherds-319dae3c92b2?source=rss-a5cd57d97916------2</link>
            <guid isPermaLink="false">https://medium.com/p/319dae3c92b2</guid>
            <category><![CDATA[ai]]></category>
            <category><![CDATA[artificial-intelligence]]></category>
            <category><![CDATA[management]]></category>
            <category><![CDATA[innovation]]></category>
            <dc:creator><![CDATA[Agnes Stenbom]]></dc:creator>
            <pubDate>Mon, 04 May 2020 07:00:18 GMT</pubDate>
            <atom:updated>2020-05-04T07:06:08.298Z</atom:updated>
            <content:encoded><![CDATA[<h3>Grand &amp; Gradual: A Call for AI Shepherds</h3><h3>The proceedings to the 1956 Dartmouth Summer Research Project on Artificial Intelligence help shed light on some key management challenges of today. What lessons on responsible AI innovation can we learn from budget lines and guest lists of the 50&#39;s?</h3><p>During the summer of 1956 some of the past century’s greatest minds in mathematics, computing and neuroscience got together for an intellectual summer camp of sorts at Dartmouth College in New Hampshire. During 8 weeks, expansive thinkers like John McCarthy, Marvin Minsky, Nathaniel Rochester and Claude Shannon (<em>yes, all men — let’s revisit that thought in a minute</em>) put their heads together to test the following hypothesis:</p><h4>“[…] every aspect of learning or any other feature of intelligence can in principle be so precisely described that a machine can be made to simulate it.”</h4><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*-DBQj2Uabu3mFrqnOAYB9A.png" /><figcaption>Photo by Margaret Minsky via <a href="http://www.achievement.org/">www.achievement.org</a></figcaption></figure><p>To be able to embark on their groundbreaking study on how to simulate human intelligence, the group behind the Dartmouth summer needed $$$.</p><p>In a <a href="http://www-formal.stanford.edu/jmc/history/dartmouth/dartmouth.html">1955 proposal </a>outlining what they wanted to achieve during their summer at Dartmouth, Minsky and crew ask for the following:</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*sTequU6A8m1o9Ol_lmDq3Q.png" /><figcaption>Example budget lines from<a href="http://www-formal.stanford.edu/jmc/history/dartmouth/dartmouth.html"> the 1955 research proposal</a></figcaption></figure><p>Would a modern day accountant feel comfortable putting that last expense line — ‘<strong>expenses for two or three people visiting for a short time</strong>’ — in the books? Questionable. Was it necessary for the Dartmouth group to have such an open-ended line when asking for financial support for their groundbreaking exploration? I think yes.</p><h3>Setting our organisations up for AI innovation</h3><p>AI has evolved since the 1950’s and now play a central role in shaping how we live, work, trade and even love. The exponential growth of both theory and practice of the past years has awaken a new breed of management challenges. How might we gear our innovation abilities to enable truly groundbreaking — and societally beneficial — AI to grow from within our organisations?</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*3TOw85LkyYwAtND2GmF0DA.png" /><figcaption>Photos through <a href="https://www.fastcompany.com/90399709/to-understand-ai-in-2019-watch-this-1960-tv-show">Fast Company</a></figcaption></figure><h4>Grand but gradual</h4><p>A key leadership mistake today is to view AI as ‘<a href="https://hbr.org/2019/07/building-the-ai-powered-organization">a plug-and-play technology with immediate returns</a>’. Simply employing cutting-edge technology is not enough, but leaders need to make sure that their company’s culture, structure, and ways of working is set up to support AI adoption and enable success.</p><p>Many business leaders expect breakthrough financial returns on their innovation endeavours, and when such fail to surface <a href="https://hbr.org/2019/12/real-innovation-requires-more-than-an-rd-budget">they cut off spending.</a> This may be related to what has been described as managers’ single-minded focus on ‘<a href="https://hbr.org/2008/02/when-small-ideas-add-up-to-som">blockbuster insights</a>’, leading them to stop projects early and hence fail to reap the benefits that could have been awarded through smaller ideas down the line.</p><p>To fully capture the potential of our new technological tools, we need to widen our horizons when looking at what they can and should be used for. <a href="https://hbr.org/2019/12/real-innovation-requires-more-than-an-rd-budget">Gina O’Connor</a> sheds light on the fact that true innovation functions should foster an expansive view of what a technology might <strong><em>become</em></strong> and then shepherd it down the most promising pathways.</p><p>Using her vocabulary, I argue that we must take greater responsibility for shepherding AI into practical use cases that benefit societies and individuals across the world. We mustn’t just follow the paths set by other, neither as individuals or organisations.</p><blockquote>AI won’t magically change the world for the better. As it’s creators, it’s our job to make sure it does.</blockquote><p>AI won’t magically change the world for the better. As it’s creators, it’s our job to make sure it does. Just like the Dartmouth group, we need to find methods for exploring grand AI ideas in gradual ways. Big statements about the groundbreaking nature of AI does little for its practical implementation.</p><h4>Small sandbox, small sandcastle</h4><p><a href="https://hbr.org/2002/08/breaking-out-of-the-innovation-box">John Wolpert </a>describes how the winds of innovation in late 90’s came to a halt in the 00’s, noting that expenses of exploring radical new business opportunities or potentially disruptive technologies went from being mandatory to ‘discretionary’. That is worrisome.</p><p>It’s not rocket science (small sandbox, small sandcastle), but the dilemma is still a tough one for leaders at this technological crossroads. Business sustainability and true innovation can co-exist, but there is no recipe dictating how. Innovators within larger organizations looking to employ AI for something truly novel might struggle to get large enough boxes to operate within to not kill their ideas.</p><p>Thinking outside the box sounds great in theory, but in practice we all want to pay rent and feed our kids.</p><h3>AI needs expansive thinkers</h3><p>For managers looking to make use of AI in ways that make sense for both people, planet and profit, understanding the need for expansive thinking is essential. Managers must enable AI workers to explore big hypotheses through small first steps, and award them space to truly explore — just like during the Dartmouth summer of 1956.</p><p>Slapping ‘moonshot’ or ‘innovation’ stickers on the laptops of AI enthusiasts won’t do the trick. This is challenge is about gradually building an organisation that is able and willing to shepherd AI innovations down their most promising pathways.</p><blockquote>Slapping ‘moonshot’ or ‘innovation’ stickers on the laptops of AI enthusiasts won’t do the trick.</blockquote><p>One action managers can take today is to offer diverse minds an opportunity to collaborate. Anyone bringing together people for an AI effort should seriously reconsider their guest list if it — like the Dartmouth project— only includes one gender, age group or ethnicity. Enabling staff to move between departments and contribute in AI projects when relevant could be key; no need to hire an army of philosophers for every new AI idea.</p><p>Another action is to allow for some open ended budget lines without clear immediate returns, 1956-style.</p><p>If we fail to offer the practical prerequisites for true exploration, <strong>we risk cutting ourselves short from the potential beneficial social and economic value AI may bring our organisations, societies and ultimately, our world.</strong></p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=319dae3c92b2" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Tech Matters: How data and digital services are helping during the COVID-19 pandemic]]></title>
            <link>https://medium.com/@agnes.stenbom/tech-matters-how-data-and-digital-services-are-helping-during-the-covid-19-pandemic-b9969d9d272e?source=rss-a5cd57d97916------2</link>
            <guid isPermaLink="false">https://medium.com/p/b9969d9d272e</guid>
            <dc:creator><![CDATA[Agnes Stenbom]]></dc:creator>
            <pubDate>Thu, 19 Mar 2020 10:57:37 GMT</pubDate>
            <atom:updated>2020-03-19T13:40:03.557Z</atom:updated>
            <content:encoded><![CDATA[<h3>How might recent technological developments help us fight a global health emergency? COVID-19 puts that question to a very real test.</h3><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*mZjKF_t6qJY1dg_Z5hj4Dg.jpeg" /><figcaption>Photo by <a href="https://unsplash.com/@visuals?utm_source=unsplash&amp;utm_medium=referral&amp;utm_content=creditCopyText">visuals</a> on <a href="https://unsplash.com/collections/9717149/coronavirus-covid-19-pandemic?utm_source=unsplash&amp;utm_medium=referral&amp;utm_content=creditCopyText">Unsplash</a></figcaption></figure><p><em>This post highlights some of the great work done by tech organisations and media workers across across the world related to the global outbreak of COVID-19. We’ll look at examples ranging from AI epidemiologists to viral TikTok dances, all working towards the common goal of providing support and relief in this new reality.</em></p><h3>Data visualizations</h3><p>When information (data) is translated into visual representations like a map or graphs it becomes easier for the human brain to understand it. For complex issues like pandemics, such simplifications are crucial in order to get important messages across to wide audiences — not just the elites of epidemiology. Let’s look at some examples!</p><ul><li><a href="https://www.washingtonpost.com/"><strong>Washington Post</strong></a>. In this great example of data journalism, Harry Stevens not just tells but <a href="https://www.washingtonpost.com/graphics/2020/world/corona-simulator/"><em>shows</em></a> readers why outbreaks like the coronavirus spread exponentially, and how we may all contribute to ‘flattening the curve’.</li></ul><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/0*pG6Kk_j-nvoY-LS9" /><figcaption>Source: <a href="https://www.washingtonpost.com/graphics/2020/world/corona-simulator/">Washington Post</a>. Visualisation retrieved 2020.03.18</figcaption></figure><ul><li><a href="https://www.bing.com/covid"><strong>Microsoft/Bing</strong></a>. By aggregating data from sources like the World Health Organization (WHO), the US Centers for Disease Control and Prevention (CDC), and the European Centre for Disease Prevention and Control (ECDC), Bing provides<a href="https://www.bing.com/covid"> a service</a> helping users understand the volume of cases on a national and global level.</li><li><a href="https://www.arcgis.com/apps/opsdashboard/index.html#/bda7594740fd40299423467b48e9ecf6"><strong>Johns Hopkin</strong>/<strong>Center for Systems Science and Engineering</strong></a> are providing a similar service, but do note the difference in their UX design. Showing the total recovered in large green letters might seem like a small move, but signalling the recovery rate could be important to getting people to trust the fact that we will get through this.</li></ul><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*KFotOPt1RCI4iCVVYY5J8g.png" /><figcaption>Source: <a href="https://www.bing.com/covid">Bing’s COVID-19 tracker</a> and Johns Hopkin’s <a href="https://www.bing.com/covid">COVID-19 Global Cases</a>. Visualisations retrieved 2020.03.19</figcaption></figure><h3>AI tracking and analysis</h3><p>When fighting a deadly virus,<a href="https://blogs.scientificamerican.com/observations/novel-coronavirus-is-a-reminder-the-best-defense-against-a-new-viral-outbreak-is-early-detection/"> early detection is crucial.</a> The speed and potential scale of impact of AI systems offers an opportunity to respond more effectively to the outbreak, and to monitor future scenarios.</p><ul><li><a href="https://bluedot.global/products/explorer"><strong>Bluedot</strong></a>. This Canadian company caught the now global virus. Their AI system, which considers more than 100 different datasets including news sources, climate data, animal populations, and airline tickets sales, <a href="https://www.cnbc.com/2020/03/03/bluedot-used-artificial-intelligence-to-predict-coronavirus-spread.html">detected an outbreak </a>of pneumonia cases in Wuhan, China on December 31, 2019 and identified the cities at the highest risk of facing outbreaks.</li></ul><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*-q3KahX9YhH4ypUuzcZP8Q.png" /><figcaption>Source: <a href="https://bluedot.global/products/explorer">Bluedot</a></figcaption></figure><ul><li><a href="https://deepmind.com/"><strong>DeepMind</strong></a>. The Alphabet subsidiary utilised their <a href="https://github.com/deepmind/deepmind-research/tree/master/alphafold_casp13">AlphaFold </a>AI system to <a href="https://deepmind.com/research/open-source/computational-predictions-of-protein-structures-associated-with-COVID-19">predict the protein structures of the coronavirus</a>. They claim not to have provided any ‘truths’ about the virus, but hope to contribute to the scientific community’s interrogation of how the virus functions.</li><li><a href="https://www.baidu.com/"><strong>Baidu</strong></a>. Much like DeepMind, the Chinese tech giant made its AI algorithm <a href="https://github.com/LinearFold/LinearFold">Linearfold</a> <a href="https://www.technologyreview.com/s/615342/how-baidu-is-bringing-ai-to-the-fight-against-coronavirus/">available for researchers</a> to help predict the virus’ structures. Sharing is caring.</li><li><a href="https://healthmap.org/en/"><strong>HealthMap.</strong></a><strong> </strong>A team of researchers working with Boston Children’s Hospital have <a href="https://www.wsj.com/articles/online-map-tracks-coronavirus-outbreak-in-real-time-11583354911">developed an AI system</a> to track the spread of virus. The system — <a href="https://healthmap.org/en/">HealthMap</a> — monitors, organizes, integrates, filters, visualices and disseminates data from social media, Google searches, blog posts, and chat rooms in nine languages. These are not the traditional sources used in epidemiologia, but could be useful for identifying early signs of an outbreak.</li></ul><h3>Video conferencing platforms</h3><p>Video conferencing platforms are allowing many of us to maintain some notion of ‘business as usual’. This new reality of social distancing, <a href="https://www.vox.com/recode/2020/3/3/21162802/tech-conferences-cancellation-coronavirus">cancellations</a> and remote work has spurred important discussions such as <a href="https://nypost.com/2020/03/16/how-to-look-good-on-video-calls-while-working-remotely-amid-coronavirus/">how to look good on a video call </a>(turns out, you should wear pants) as we all pull our weight in flattening the curve.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/800/0*Ir03yv2E69smTo8n" /><figcaption>Source: S<a href="https://www.sciencealert.com/dragging-out-the-coronavirus-epidemic-is-important-in-saving-lives">cience Alert / Matthew McQueen</a></figcaption></figure><ul><li><a href="https://zoom.us/"><strong>Zoom</strong></a>. By <a href="https://zoom.us/docs/en-us/covid19.html">sharing information and advice</a> regarding remote working and educating, the team behind Zoom wants to ‘help you most effectively use Zoom as we all navigate the coronavirus pandemic’. The video conferencing tool made it to #1 among the <a href="https://www.vox.com/recode/2020/3/11/21173449/microsoft-google-zoom-slack-increased-demand-free-work-from-home-software">most downloaded business app</a>s in the U.S. on iOS last week.</li><li><a href="https://www.loom.com/"><strong>Loom</strong></a>. The video recording and sharing service <a href="https://www.forbes.com/sites/shamahyder/2020/03/15/coronavirus-champions-a-running-list-of-brands-getting-it-right/">has made Loom Pro free</a> for teachers and students at K-12 schools, universities, and educational institutions. They’ve also removed the recording limit on free plans and cut the price for Loom Pro in half. In a <a href="https://techcrunch.com/2019/02/19/loom-video/">2019 interview,</a> co-founder Shahed Khan described their product as an effort to<em> </em>‘make it very easy to bring humanity back’ into meetings. Their response to the COVID-19 pandemic sure is humane (and probably a smart business move, too).</li></ul><h3>Social media</h3><p>During the past years, social media and platforms for user-generated content have often been described as <a href="https://www.nytimes.com/2019/02/19/technology/youtube-conspiracy-stars.html">antisocial forces</a> awarding <a href="https://www.morningbrew.com/emerging-tech/stories/2020/03/18/facebook-google-microsoft-twitter-reddit-team-coronavirus-misinformation">viral wings</a> to fraudulent information. But this pandemic is bringing out hope for the good old promise of the internet: <a href="https://www.nytimes.com/2020/03/15/business/media/coronavirus-facebook-twitter-social-media.html">democratization of information through organized communities.</a></p><ul><li><a href="https://www.pinterest.com/"><strong>Pinterest</strong></a><strong>. </strong>The platform’s strict <a href="https://help.pinterest.com/en/article/health-misinformation">health misinformation policy</a> outlines some basic house rules like not allowing users to save content that includes advice where there may be immediate and detrimental effects on a Pinner’s health or on public safety. Anyone <a href="https://www.buzzfeednews.com/article/craigsilverman/pinterest-is-blocking-coronavirus-searches-and-people-are">searching for “COVID-19” or “coronavirus”</a> are only shown content from internationally recognised health organisations.</li></ul><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/0*I453D5Dgz1l_ldw9" /><figcaption>Source: <a href="https://www.pinterest.com/">Pinterest</a></figcaption></figure><ul><li><a href="https://twitter.com/home"><strong>Twitter</strong></a>. As noted by Joanna Geary, senior director of curation at Twitter, in a <a href="https://www.nytimes.com/2020/03/15/business/media/coronavirus-facebook-twitter-social-media.html">NYT interview</a> (2020/03/17), ‘the people who are part of professional communities across epidemiology and other kinds of public health — normally they’re talking to each other, but right now they’re talking to us all’. As described in the interview, Geary’s team gathers compelling and informative tweets about COVID-19 and shows them at the top of all users’ feeds. This earns the posts higher engagement, which in turn signals to Twitter’s algorithm that more people should see them.</li></ul><iframe src="https://cdn.embedly.com/widgets/media.html?src=https%3A%2F%2Fwww.youtube.com%2Fembed%2FBtulL3oArQw%3Ffeature%3Doembed&amp;display_name=YouTube&amp;url=https%3A%2F%2Fwww.youtube.com%2Fwatch%3Fv%3DBtulL3oArQw&amp;image=https%3A%2F%2Fi.ytimg.com%2Fvi%2FBtulL3oArQw%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/28b1851661a039fb879d3cd7a39dc64e/href">https://medium.com/media/28b1851661a039fb879d3cd7a39dc64e/href</a></iframe><ul><li><strong>TikTok x Vietnamese PSA</strong> The undeniably catchy <a href="https://www.youtube.com/watch?v=BtulL3oArQw">coronavirus PSA song </a>by Vietnam’s National Institute of Occupational Safety and Health motivated dancer Quang Đăng’s to launch <a href="https://www.billboard.com/articles/news/international/9329174/vietnam-coronavirus-psa-song-best-tiktok-dance-videos">the TikTok dance challenge</a> #GhenCoVyChallenge. His dance transforms hand-washing techniques into groovy dance moves, and yes, o<a href="https://www.bbc.co.uk/newsround/51791722">f course it went viral.</a></li><li><strong>TikTok x World Health Organisation. </strong>WHO also turned to<a href="https://www.tiktok.com/@who"> TikTok </a>in an attempt to boost the sharing of accurate information about the virus and its spread. They are perhaps not as down with the kids as Quang Đăng, but at least they’ve launched a hashtag (#SafeHandsChallenge).</li></ul><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*xq3ooZwrVQbUHCwKF3rNgw.png" /><figcaption>Source: <a href="https://www.tiktok.com/@who">TikTok / WHO</a></figcaption></figure><h3>EdTech</h3><p>Social distancing and quarantines have already led to increased consumption of online learning content as people across the world look to use this time for self-improvement.</p><ul><li><a href="https://www.adobe.com/"><strong>Adobe</strong></a> is <a href="https://www.forbes.com/sites/shamahyder/2020/03/15/coronavirus-champions-a-running-list-of-brands-getting-it-right/">providing</a> higher education and K-12 institutional customers the opportunity to add temporary “at-home” access for students and educators <a href="https://www.forbes.com/sites/shamahyder/2020/03/15/coronavirus-champions-a-running-list-of-brands-getting-it-right/">through May 31</a> at no additional cost.</li><li><a href="https://www.linkedin.com/feed/"><strong>LinkedIn</strong></a> is <a href="https://www.forbes.com/sites/shamahyder/2020/03/15/coronavirus-champions-a-running-list-of-brands-getting-it-right/">opening up</a> 16 of its learning courses for free. The courses include tips on productivity, remote relationships, virtual meeting tools and work-life balance.</li><li><a href="https://kahoot.com/"><strong>Kahoot</strong></a>. The game-based learning platform Kahoot is <a href="https://kahoot.com/blog/2020/02/27/kahoot-free-access-schools-higher-education-coronavirus/">giving away free access</a> to its Premium offering to schools and universities impacted by COVID-19 closures. With the Kahoot tooling for distance learning, teachers can host gamified learning sessions via video and assign students games for homework.</li></ul><h3>What’s next?</h3><p>Our ways of working, shopping, socialising and learning are changing rapidly through this global crisis. As we adapt to this new reality, we need to be prepared for the potential lasting changes our current behaviours might spur.</p><blockquote>Are these the days we’ll look back on as a catalyst for a truly digital-first world?</blockquote><p>Are these the days we’ll look back on as the catalyst for a truly digital-first world? Will we shift to a<a href="https://skift.com/2020/03/17/after-the-virus-10-consumer-trends-for-a-post-coronavirus-world/?mntr_id=GGmWmB"> virtual experience economy</a>? Will e-commerce prevail as the new standard across categories? Will open source and cross-organisational collaborations become more prevalent as workers connect more to their tasks than their physical organisation?</p><p>One notable aspect of COVID-19 is that <strong>everyone</strong> are pushed to digital tools and services— including elders. ‘Digital laggards’ are suddenly entering a brave new world of technology, and as organisations we need to be agile and responsive to potential changes in consumer behaviours and expectations.</p><h4>Don’t forget about digital ethics</h4><p>In our attempts to flatten the curve and provide support during this global crisis, let’s not forget about important principles such as the importance of ensuring accuracy, respecting privacy, protecting minorities, or safeguarding transparency. If you’re working on a product or service aimed at supporting the crisis management, please ask health care (and preferably pandemic) experts for feedback before launch.</p><p><strong>Finally, thank you to everyone in the tech community contributing with innovation and optimism this challenging situation!</strong></p><p><em>Do you have suggestions on efforts that should be added to this post? Send me a DM!</em></p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=b9969d9d272e" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Why Preventing Algorithmic Bias is in Your Media Business’ Interest]]></title>
            <link>https://medium.com/@agnes.stenbom/why-preventing-algorithmic-bias-is-in-your-media-business-interest-d750e6f06d78?source=rss-a5cd57d97916------2</link>
            <guid isPermaLink="false">https://medium.com/p/d750e6f06d78</guid>
            <category><![CDATA[technology-trends]]></category>
            <category><![CDATA[bias]]></category>
            <category><![CDATA[society]]></category>
            <category><![CDATA[artificial-intelligence]]></category>
            <category><![CDATA[ai]]></category>
            <dc:creator><![CDATA[Agnes Stenbom]]></dc:creator>
            <pubDate>Thu, 06 Dec 2018 07:18:22 GMT</pubDate>
            <atom:updated>2019-01-02T13:18:32.622Z</atom:updated>
            <content:encoded><![CDATA[<p><em>Media businesses looking to serve the interest of diverse societies need to reflect on how we implement new technologies. An AI is no smarter than the data it eats, and it is in our ethical and financial interest to think twice about what we put on its plate!</em></p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*SJ4IZGQlLH0-Inm_62YQQQ.jpeg" /><figcaption>@danielkchung</figcaption></figure><p>You have probably heard about <em>unconscious bias </em>before. From keeping us out of danger to telling genders apart, our biases kick in automatically to help our brains process billions of information bits. Every second.</p><p>Unconscious stereotyping around e.g. appearances, beliefs or nationalities is part of being human, yet the promise of Artificial Intelligence (AI) has awoken hope for a less biased future. Could truly objective information processes be around the corner?</p><h4>Say bye-bye to your dreams of objective algorithms</h4><p>AI is an umbrella term which refers to computer systems that reasons and adapts based on algorithms (rules) and data. Many seem to have faith in the neutrality of AI, but algorithmic systems do not<strong> </strong>live in a vacuum. They are socio-technical creations, built and used by human beings. As we trust AI with more of our lives’ processes — ranging from credit scores to healthcare — awareness of our human impact becomes increasingly important.</p><p>Connected to unconscious bias, <strong><em>algorithmic bias</em></strong> occurs when the implicit values of those collecting and employing data slips into computer systems. Or in other words: when we build our human biases into AI.</p><h4><strong><em>Example: Gender-biased translation services</em></strong></h4><blockquote>A translation AI has to be fed with training data in order to learn that one word represents or is connected to another in different languages. That data is not always easy to get hold of — you need large quantities of professionally translated text, which is both expensive and time consuming. Companies therefore often turn to old archives of official documents. These tend to represent gender roles of the past, when e.g. only men worked as doctors and women worked domestically.<strong> </strong>As the algorithm (“the rules of the AI”) sets out to match the most frequent pronoun with any given word, traditionally male words like ‘doctor’ or ‘professor’ are paired with the pronoun ‘he’ when translated from gender neutral languages.</blockquote><p>MIT Media Lab’s Joy Buolamwini showed us how<a href="http://gendershades.org/"> image recognition services mainly work for white people</a>, <a href="https://www.reuters.com/article/us-amazon-com-jobs-automation-insight/amazon-scraps-secret-ai-recruiting-tool-that-showed-bias-against-women-idUSKCN1MK08G">tech giants are struggling to find diverse talents due to biased recruitment AI:s</a> and <a href="https://www.technologyreview.com/s/601111/why-microsoft-accidentally-unleashed-a-neo-nazi-sexbot/">corporate chat-bots turn misogynic</a>. The list of algorithmic bias brought to market goes on and on.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*OxHwGtu-p9bRFHYRl-5jYw.jpeg" /><figcaption>@danielkchung</figcaption></figure><h4>Can a desperate industry help prevent it?</h4><p>In a<a href="https://tinius.com/2018/11/11/interview-with-ben-thompson/"> recent interview on the future of media,</a> Stratechery founder Ben Thompson noted that ethical collapses happen in desperation. Is the AI industry desperate? Well, some of it might be. The development of AI is often depicted as a modern arms race between China and the US. While the US has a tradition of technological innovation, the country is now struggling to keep up with Chinese progress.</p><p>If data is the new oil, China’s giant online population makes it the owner of the hottest natural resource of the future: human data. As China lacks the privacy restrictions of many western countries, the eastern giant is able to take leaps forward by performing<a href="https://www.nytimes.com/2018/07/08/business/china-surveillance-technology.html"> large scale societal experiments</a> with AI.</p><p>In the west we tend to accredit negative development trends in AI to China. But reality is that American and European companies are<a href="https://euobserver.com/opinion/142056"> running so fast to keep up</a> that we tend to forget about some basic components of service design — as ethics.</p><h4>Ethics will be a commercial win</h4><p>Ethics will have increased business importance going forward. Anna Felländer, founder of AI Sustainability Center, notes that ethical AI will become a<a href="https://tinius.com/2018/01/15/ai-sustainability-a-new-form-of-csr/"> ‘moral license to operate’</a> as global audiences increasingly demand transparency and fair practices in tech. Felländer concludes that:</p><blockquote>“Scandals due to misuse of data, inconsistent or unethical AI, machine bias or immature AI will make the break of customer trust more painful.”</blockquote><p>Consumer trust will be essential as data becomes the commodity of the future. Failing to deliver ethical AI may break the precious bonds businesses have with consumers, and as pointed out by Felländer the tolerance level for mistakes will get exponentially lower as AI grows in application.</p><p>Ethics will also become a talent acquisition- and retention topic in the future. The next generation to enter the workforce, Generation Z, are<a href="https://www.huffingtonpost.com/entry/think-millennials-are-purpose-driven-meet-generation_us_5a1da9f3e4b04f26e4ba9499"> being called ‘the Purpose Generation’</a>. Attracting talents will in many cases be a game of (you guessed it!) ethics.</p><h4>… but we’re not there just yet</h4><p>While many companies join the call for increased diversity and ethical practices in AI, the tech industry is facing problems. Want numbers?<a href="https://www.forbes.com/sites/lisawinning/2018/03/13/its-time-to-prioritize-diversity-across-tech/#4f72ce2416f8"> 95% of the global tech workforce is white</a>,<a href="https://2018.stateofeuropeantech.com/"> 93% of all funds raised by VC-backed tech companies in Europe went to all-male founding teams in 2018</a>, and only<a href="https://www.wired.com/story/artificial-intelligence-researchers-gender-imbalance/"> 12% of the global AI workforce is female</a>. Remember how algorithmic bias happens? Historical (or present…) data does not always represent the future we wish to achieve.</p><p>Niche media products are on the rise and as we fuel our brands through technology we will need parts of our biases to steer the way. The pressing action here is not to eliminate all kinds of bias (they are part of humanity, after all), but to reflect on how the data and algorithms we use may affect our users. Failing to do so can have both ethical and financial costs in the future.</p><p>There is no one answer on how to prevent algorithmic bias, but we need a dialogue within and among media businesses. Please reach out should you and your team be interested in furthering the discussion!</p><p><strong>Agnes Stenbom<br></strong>Technology Trend Analyst<br>Schibsted Media Group</p><p>— — — — — — — — — — — — — — — — — — — — — — — — — — — — — —</p><p><strong><em>Update 09–12–18:</em></strong><em> Good news! Google </em><a href="https://l.messenger.com/l.php?u=https%3A%2F%2Ftechcrunch.com%2F2018%2F12%2F07%2Fgoogle-translate-gets-rid-of-some-gender-biases%2F&amp;h=AT3i95Q42btvDF5jlJHr8yRQe7JUf83BbtuNuMMuhnM6sZ-HXJAJEUyFYUbPCX5TwscRo1SAspqmXbxQ41YtBfRWz-2z9MSqlc2vjFZwvkvVytahH46l3fjHacJkwjW6dI0ilgw"><em>just announced</em></a><em> updates in their algorithm to fight gender bias in translation.</em></p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=d750e6f06d78" width="1" height="1" alt="">]]></content:encoded>
        </item>
    </channel>
</rss>