<?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[Humane AI - Medium]]></title>
        <description><![CDATA[Humane AI is a newsletter about social and human rights implications  of Artificial Intelligence. Subscribe here: https://www.royapakzad.co/newsletter/. - Medium]]></description>
        <link>https://medium.com/humane-ai?source=rss----7ad53ef01cf3---4</link>
        <image>
            <url>https://cdn-images-1.medium.com/proxy/1*TGH72Nnw24QL3iV9IOm4VA.png</url>
            <title>Humane AI - Medium</title>
            <link>https://medium.com/humane-ai?source=rss----7ad53ef01cf3---4</link>
        </image>
        <generator>Medium</generator>
        <lastBuildDate>Thu, 19 Nov 2026 08:54:36 GMT</lastBuildDate>
        <atom:link href="https://medium.com/feed/humane-ai" rel="self" type="application/rss+xml"/>
        <webMaster><![CDATA[yourfriends@medium.com]]></webMaster>
        <atom:link href="http://medium.superfeedr.com/" rel="hub"/>
        <item>
            <title><![CDATA[From polygraphs to modern lie detectors: what past failures can teach us today]]></title>
            <link>https://medium.com/humane-ai/from-polygraphs-to-modern-lie-detectors-do-they-keep-our-borders-safe-a666c5890e98?source=rss----7ad53ef01cf3---4</link>
            <guid isPermaLink="false">https://medium.com/p/a666c5890e98</guid>
            <category><![CDATA[borders]]></category>
            <category><![CDATA[human-rights]]></category>
            <category><![CDATA[traveling]]></category>
            <category><![CDATA[machine-learning]]></category>
            <category><![CDATA[ai]]></category>
            <dc:creator><![CDATA[Roya Pakzad]]></dc:creator>
            <pubDate>Sat, 07 Sep 2019 00:23:18 GMT</pubDate>
            <atom:updated>2019-09-12T20:58:31.481Z</atom:updated>
            <content:encoded><![CDATA[<h4>From the 15th edition of my newsletter, <a href="https://mailchi.mp/b15020724d83/humane-ai-newsletter-15">Humane AI</a></h4><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*6-VtcLmwRmcPh24Tzs2b6g.jpeg" /><figcaption>El Paso Border Patrol agents &amp; a group of Guatemalan migrants. (photo from U.S. Customs and Border Protection/U.S. Customs and Border Protection, <a href="https://www.washingtonpost.com/graphics/2019/national/a-look-at-why-a-wall-wont-stop-immigration-surge/">source</a>)</figcaption></figure><p>Summer with all its traveling is over! This issue of my newsletter, Humane AI is about traveling, inspired by my summer vacation. But, more importantly, writing this issue is inspired by a passport-control officer in an airport in Varna, Bulgaria, skeptically scratching my Schengen visa with the edge of a coin, to make sure that the visa sitting confidently in my Iranian passport is actually real!</p><p>For the next few minutes, I invite you to read about AI being used in immigration and traveling, geo-tagging and its negative impacts on the environment, Airbnb’s “Experiences” and more.</p><p><strong>✏ </strong><a href="https://theintercept.com/2019/07/26/europe-border-control-ai-lie-detector/"><strong>From polygraphs to AI-enabled lie detectors: they all lie but are still in use. This time, for travelers! </strong></a><br>“Take a quarter out of your pocket. Flip it, if it’s heads, you are lying; if it’s tails, you are telling the truth… That’s exactly the same degree of accuracy that this polygraph instrument has.” <a href="https://www.youtube.com/watch?v=N3fHkCFxgQQ">Said</a> by Doug Williams, a former Oklahoma City police detective who ran polygraph examinations between 1972 to 1979. Later, he decided to spend his life training people to pass the test via different techniques including calming themselves by picturing being on the beach, thinking about their greatest fears, or even clenching their anus!</p><p>Despite the scientific community’s concerns about the lack of accuracy of these tools, law enforcement agencies still use polygraphs for finding criminals, spies, etc. According to <a href="https://theintercept.com/2019/07/26/europe-border-control-ai-lie-detector/">this <em>Intercept</em> article</a>, an AI-enabled lie-detector — a virtual policeman piloted at the Hungary, Greece and Latvia airports — is there to ask you questions at the border. “You provide your answers verbally, and the virtual policeman uses your webcam to scan your face and eye movements for signs of lying.” The project was developed by researchers at Manchester Metropolitan University, “who say that the technology can pick up on ‘micro gestures’ a person makes while answering questions on their computer, analyzing their facial expressions, gaze, and posture.” The article raises concerns about the accuracy and transparency of this tool. To learn more about the study involved, read <a href="https://www.documentcloud.org/documents/6215340-Intelligent-Deception-Detection-Through-Machine.html">this paper</a>. The number of people who participated in the study to train and test this tool is 32; the overall accuracy, 75%; and the funding supplied by the EU, 4.5 million euros!</p><p>📎 If you want to dig into studies about using machine learning algorithms to infer one’s well-being, general intelligence, writing ability, verbal fluency, sexual orientation, political orientation, personality traits, autism, depression, dyslexia, psychopathy, and stress, check out this paper entitled “<a href="https://link.springer.com/article/10.1007/s11023-019-09497-4">Can Machines Read our Minds?</a> “ published by the Oxford Internet Institute. Researchers examined 26 studies on the possibility of predicting psychological traits and mental states of individuals via their social media activities, facial features, etc. The paper also raises questions about the issue of trust, informed consent, and digital identity.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*HUPaQJi8OxThAa2BFEq4hw.jpeg" /><figcaption><em>An advertisement for the Keeler Polygraph, mid 1950s. Via UC Santa Cruz Special Collections.</em></figcaption></figure><p><strong>✏</strong><a href="https://www.brennancenter.org/publication/social-media-monitoring"><strong>How the U.S. Department of Homeland Security uses digital data in the name of national security</strong></a><br>The idea of monitoring social media profiles to grant an entry visa to the U.S. is not a new thing (see this <a href="https://www.brennancenter.org/analysis/timeline-social-media-monitoring-vetting-department-homeland-security-and-state-department">timeline</a>). However, last week, social media monitoring took an Orwellian turn by learning that a Palestinian freshman Harvard student was deported because of his friend’s facebook post. Read the story <a href="https://www.theverge.com/2019/8/31/20837448/social-media-dhs-cbp-surveillance-us-border-ismail-ajjawi-harvard">here</a>. The authors of <a href="https://www.brennancenter.org/publication/social-media-monitoring">the paper, Social Media Monitoring</a>, questioned the effectiveness of collecting social media data for the sake of border security. Also, according to the report, “DHS’s social media monitoring pilot programs seem to have focused in large part on Muslims: at least two targeted Syrian refugees, one targeted both Syrian and Iraqi refugees, and the analytical tool used in at least two pilots was tailored to Arabic speakers.” According to the report, one of the impacts of such initiatives has been “self-censorship of people applying for visas as well as their family members and friends.”</p><p>📎 For those of you interested in law, I suggest reading this great Lawfare article entitled “<a href="https://www.lawfareblog.com/national-vetting-enterprise-artificial-intelligence-and-immigration-enforcement">The National Vetting Enterprise: Artificial Intelligence and Immigration Enforcement</a>” by Chinmayi Sharma. The article goes into the details of the Department for Homeland Security proposals for using AI in immigration. It breaks down the specific roles that USCIS, ICE, and CBP play in executing DHS orders.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*KaepvKyloY5Zt5GwLASi-Q.jpeg" /><figcaption><em>Me, my mother, and my sister laughing in Malaysia.</em></figcaption></figure><p>Why is this so important to me? I’ll tell you: because of the nature of my work (relating to human rights implications of new technologies) I have decided not to travel back to my country in order to avoid any kind of potential <a href="https://www.hrw.org/news/2018/09/26/iran-targeting-dual-citizens-foreigners">governmental hostility toward dual citizens</a>. For fourteen days out of all these 365 days in a year, my mother, my sister and I try to re-unite somewhere outside Iran. We plan months before, get excited, go through visa checklists and interviews over video calls. But then I see them getting rejected again and again, barring them from entering the U.S. (due to the Muslim Ban) and all European countries in the Schengen zone. Sometimes, I wonder what is the value of all these “reliable vetting systems” and “safe border technologies” if, in the end, their collective decision is to ban this: the joy of being with your family and being <strong>safe</strong>.</p><p><strong>✏</strong><a href="https://www.nytimes.com/2018/10/12/world/europe/seven-sisters-cliffs-tourism-pictures.html"><strong>How geotagging on Instagram has been ruining environment — and taking human lives?!</strong></a><br>“When we search for London on social media, it’s the first thing we see,” said a South Korean visitor to the Seven Sisters cliffs in England. Because of social media geo-tagging, influencer posts, and search algorithms, the Seven Sister cliff — formerly known only to avid hikers — has become one of the favorite spots for tourism. According to the article “the growing number of visitors has raised concerns about the risks of having so many people high up on the cliffs. Last year, Kim Hye-won, a Korean student, died after losing her footing there.” According to <a href="https://www.change.org/p/encouraging-the-social-media-generation-to-behave-more-responsibly-outdoors">this Change.org petition</a>, “when photos of a location gain popularity on Instagram (via the geotag and hashtag features, plus shares or ‘features’), the result is often a surge in visitation. Unfortunately, a number of outdoor destinations all over the world have been closed or destroyed due to the negative impacts from this influx of visitors.”</p><p>Signatories of <a href="https://www.change.org/p/encouraging-the-social-media-generation-to-behave-more-responsibly-outdoors">the Change.org petition</a> ask Facebook/Instagram to allow users to “report violations that are both illegal and harmful to the environment” and implement a better content moderation practices to curb the virality of photos/accounts which promote harmful practices.</p><p><strong>✏</strong><a href="https://medium.com/airbnb-engineering/machine-learning-powered-search-ranking-of-airbnb-experiences-110b4b1a0789"><strong>Machine Learning-powered search ranking of Airbnb Experiences</strong></a><br>This summer, during my trip, I noticed that Airbnb has started a service called <a href="https://www.airbnb.com/help/article/1581/what-are-airbnb-experiences">Experience</a>s to help locals teach classes to guests or take them on local tours. <a href="https://medium.com/airbnb-engineering/machine-learning-powered-search-ranking-of-airbnb-experiences-110b4b1a0789">This post</a> published on the Airbnb Engineering and Data Science Medium page goes into details about Airbnb’s plan to better personalize this service. Here is an excerpt: “We can use the country information to personalize the Experience ranking based on Categories preferred by users from those countries. For example, historical data tells us that when visiting Paris Japanese travelers prefer Classes &amp; Workshops (e.g. Perfume making), US travelers prefer Food &amp; Drink Experiences, while French travelers prefer History &amp; Volunteering. We used this information to engineer several personalization features at the Origin — Destination level.”</p><p>Okay, I’ll skip over expressing my speculation about this service’s potential discriminatory impacts because of training on biased historical data, or the effects on local businesses, or potential concerns about treating different demographics in different ways — note that <a href="https://atlantablackstar.com/2019/08/19/black-oregon-women-score-undisclosed-settlement-in-discrimination-suit-against-airbnb/">Airbnb has already been struggling to address</a> racial discriminations in its services. I won’t even go there; I just want to highlight this sentence from the Medium post: “<em>US travelers</em> prefer<em> Food &amp; Drink </em>Experiences, while <em>French travelers</em> prefer <em>History &amp; Volunteering?!! </em>This made me burst out laughing! What can I say my American friends? ¯\_(ツ)_/¯ I just know that this sentence, all by itself, has the potential to trigger a new “<a href="https://en.wikipedia.org/wiki/Freedom_fries">Freedom Fries</a>” proposal!</p><p><em>This has been an excerpt from the </em><a href="https://mailchi.mp/b15020724d83/humane-ai-newsletter-15"><em>Humane AI newsletter #15</em></a><em>, run by me, </em><a href="https://www.royapakzad.co/"><em>Roya Pakzad</em></a><em>. You can subscribe to the newsletter </em><a href="https://royapakzad.us17.list-manage.com/subscribe?u=9138308bb26620c53a0881c20&amp;id=8152ed9f0c"><strong><em>here</em></strong></a><strong><em>.</em></strong></p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=a666c5890e98" width="1" height="1" alt=""><hr><p><a href="https://medium.com/humane-ai/from-polygraphs-to-modern-lie-detectors-do-they-keep-our-borders-safe-a666c5890e98">From polygraphs to modern lie detectors: what past failures can teach us today</a> was originally published in <a href="https://medium.com/humane-ai">Humane AI</a> on Medium, where people are continuing the conversation by highlighting and responding to this story.</p>]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[When AI Meets Nowruz]]></title>
            <link>https://medium.com/humane-ai/when-ai-meets-nowruz-b249274370ba?source=rss----7ad53ef01cf3---4</link>
            <guid isPermaLink="false">https://medium.com/p/b249274370ba</guid>
            <category><![CDATA[human-rights]]></category>
            <category><![CDATA[ai]]></category>
            <category><![CDATA[machine-learning]]></category>
            <category><![CDATA[nowruz]]></category>
            <dc:creator><![CDATA[Roya Pakzad]]></dc:creator>
            <pubDate>Mon, 15 Apr 2019 05:46:34 GMT</pubDate>
            <atom:updated>2019-04-15T05:46:33.890Z</atom:updated>
            <content:encoded><![CDATA[<h4>From the 14th edition of my newsletter, <a href="https://mailchi.mp/ce30aaadc443/humane-ai-newsletter-14">Humane AI</a></h4><figure><img alt="" src="https://cdn-images-1.medium.com/max/72/0*pTFdqy_LaFCXevPT.png" /></figure><p>We Iranians, along with more than 200 million others — mostly from Eurasian countries — celebrate Nowruz, the first day of Spring, and the beginning of the Persian New Year. This is a special edition of Humane AI covering AI in countries which celebrate Nowruz and anything related to their culture and languages. Read on to learn about Natural Language Processing in Farsi, the Digital Silk Road, early warning systems, AI and Nowruz, and more. Nowruz Mobarak!</p><p><strong>✏</strong><a href="https://openai.com/blog/better-language-models/"><strong>How does OpenAI’s language model describe Nowruz?</strong></a><br>Here is the text that I entered into the GPT-2 language model: “Nowruz is the first day of Spring. The exact moment of Spring Equinox. It’s when Iranians, Uzbeks, Afghan, Tajiks celebrate and start their new year.” And here’s how it continued [GPT-2 🤖starting] Nowruz means that we celebrate with joy because of spring and because we want to make sure that we are happy. The first time I saw Iranians celebrating was in the village of Nangarhar. I was very surprised to see that they were celebrating with fireworks and singing songs and dancing. So I thought that maybe they were just trying to show off.” I used <a href="https://gpt2.apps.allenai.org/?text=Joel%20is">GPT-2 Explorer by the Allen Institute for AI</a> to make this. It’s fun, play around.</p><p>GPT-2 is OpenAI’s latest language model. It has a simple goal: “predicting the next word.” OpenAI decided not to release the full trained model. They have a legitimate concern that this model has a potential to be used by malicious actors to “generate misleading news articles, impersonate others online, automate the production of abusive or faked content to post on social media.” But their decision created controversy due to the fact that AI research communities — all around the world — have benefited significantly from the culture of open research. In <a href="https://www.partnershiponai.org/when-is-it-appropriate-to-publish-high-stakes-ai-research/">this blog post</a>, Partnership on AI reflects on the culture of openness in the AI community and the possibility of setting up community-wide norms: “There <em>is not </em>a consensus on proposed AI research review norms… However, there <em>is</em> a consensus that norms and review parameters should be standardized across the AI community if the community chooses to restrict or reduce research openness (in contrast to organizations independently designing their own review processes).”</p><p>+ This talk of standardization, process, review, and moving slowly reminded me of <a href="https://aeon.co/essays/most-of-the-time-innovators-don-t-move-fast-and-break-things">this amazing piece</a> by W Patrick McCray, a historian of technology:</p><p>“At the same time, when we step out of the shadow of innovation, we get new insights about the nature of technological change. By taking this broader perspective, we start to see the complexity of that change in new ways. It’s then we notice the persistent layering of older technologies […] We start to see the intangibles — the standards and ideologies that help to create and order technology systems, making them work at least most of the time. We start to see that technological change does not demand that we move fast and break things. Understanding the role that standards, ideologies, institutions — the non-thing aspects of technology — play, makes it possible to see how technological change actually happens, and who makes it happen.”</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1000/0*AXBhqRpq7c-0MAdk.jpg" /><figcaption><a href="https://en.wikipedia.org/wiki/American_Progress"><em>American Progress by John Gast (1872)</em></a></figcaption></figure><p><strong>✏</strong><a href="https://www.blog.google/products/search/helping-keep-people-safe-ai-enabled-flood-forecasting/"><strong>Keeping people safe with AI-enabled flood forecasting</strong></a><br>In the past month, right at the height of the Nowruz celebrations, many cities in Iran suffered massive <a href="https://www.nytimes.com/2019/03/26/world/middleeast/iran-floods-dead.html">flash floods</a>. It was heartbreaking. Are early warning systems mature enough to help mitigate the damages of natural disasters? Google has been working on developing ML models for flood forecasting and crisis response. According to a recent <a href="https://www.blog.google/products/search/helping-keep-people-safe-ai-enabled-flood-forecasting/">blog post</a> and <a href="https://ai.googleblog.com/2019/03/a-summary-of-google-flood-forecasting.html">workshop</a>, Google is using “AI and significant computational power to create better forecasting models that predict when and where floods will occur and incorporating that information into <a href="https://google.org/publicalerts/map?hl=en&amp;gl=US">Google Public Alerts</a>. A variety of elements — from historical events to river level readings, to the terrain and elevation of a specific area — feed into [their] models. From there, [they] generate maps and run up to hundreds of thousands of simulations in each location.” However, neither <a href="https://www.youtube.com/watch?v=jEHER63hfl8">Google Public Alerts</a> nor <a href="https://www.blog.google/products/search/helping-people-crisis/">SOS Alerts</a> seems to be functional in Iran.</p><p>+ On a related topic, I’ve been collecting Tweet IDs relating to the recent flash floods in Iran. My goal is to conduct a research project on information control during times of crisis. Please get in touch if you are interested in getting access to the dataset or interested in collaborating.</p><p><strong>✏“Farsi Shekar Ast”: Natural Language Processing in Farsi</strong><br>We Iranians hold the language of Rumi, Khayyam, and Hafiz close to our hearts, as exemplified by a saying, “Farsi Shekar Ast,” or <em>Farsi is sugar!</em> As I watch my fellow Iranians “caramelizing” one another on social media, I’ve been thinking a lot about the challenges of automated content moderation in Farsi. How is content moderation done in the context of addressing hate speech and disinformation? This led me to look into the state of NLP and, in particular, sentiment analysis in Farsi. <a href="https://github.com/users/royapakzad/projects/1">Here</a> I listed some of the available datasets and crowd-sourcing projects (If you are working on this topic or you know resources, please let me know.)</p><p>In addition, this is a <a href="https://arxiv.org/pdf/1810.11383.pdf">great paper</a> listing challenges of Natural Language Processing in East African languages. Some challenges include: handling rapid code-switching (switch between languages) with models trained on single language corpora, difficulties of location extraction from Natural Language, different accents, and the idiomatic nature of East African English.</p><p><strong>✏</strong><a href="http://www.ponarseurasia.org/memo/chinese-artificial-intelligence-projects-expand-eurasian-cities"><strong>Eurasian countries, the Digital Silk Road, and “smart” cities</strong></a><br>“As a global leader in developing and using surveillance technologies, China is exporting its high-tech brand of authoritarianism to its neighbors in Eurasia. All of the systems are part of the ‘smart city’ and ‘safe city’ concepts that have been gaining popularity around the world as a response to rapid urbanization. Closed-circuit television cameras and facial recognition technologies are particularly sought products. Chinese tech companies have the strongest position in Eurasian markets in these types of capabilities thanks to their regional presence, long-term credit structures, and a high tolerance for financial risk. Eurasian countries looking to leapfrog into the digital future are adopting virtual authoritarianism, with a lack of oversight legislation, which risks suppressing political dissent and further marginalizing already disadvantaged groups.”</p><p>+ Here in the Bay Area, the <a href="https://strushucb.github.io/opp/">Bay Area Surveillance Project</a> looks into different smart city technologies which have raised concerns regarding privacy and equal protection. Technologies such as CCTV camera, Automated License Plate Recognition (ALPR), Shotspotter, Cell Site Simulator (Stingrays), and social media monitoring have proven to further surveillance. Check out this very interesting interactive<a href="https://strushucb.github.io/opp/"> website</a> to see “how each system can be used by organizations from the city level — for example, Oakland — to the federal level like DHS’ agencies.” You should also read Ben Green’s <a href="https://mitpress.mit.edu/books/smart-enough-city">The Smart <em>Enough</em> City</a>.</p><p><strong>✏</strong><a href="https://www.nytimes.com/2019/03/30/world/asia/afghan-girls-robotics-team-fatemah-qaderyan.html"><strong>Afghanistan’s all-girl robotics team and the future of STEM in the country</strong></a><br>In Afghanistan, the all-girl robotics team has been a source of inspiration for many. <a href="https://www.nytimes.com/2019/03/30/world/asia/afghan-girls-robotics-team-fatemah-qaderyan.html">This <em>NYT</em> article</a> profiled their journey to major high-school robotics competitions and their future plan to start a robotics club in Afghanistan. “Fatemah [the captain of the robotics team] and Ms. Mahboob [Founder and CEO of Afghan Citadel Software Company] met with Afghanistan’s president, Ashraf Ghani, about a plan to start a robotics club in Kabul, as well as a technology-focused school, which the Yale Engineering Department has offered to design.” In addition, through a partnership with Ada-AI and Digital Citizen Fund, they have started a crowd-funding campaign to “build a high school in Afghanistan that provides students with critical technical skills and exposure to emerging technologies and AI.” Read more <a href="https://ada-ai.org/projects/">here</a>.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/820/1*_0t2tvAb9GPz3h0J2zsgOA.jpeg" /><figcaption><a href="https://www.loc.gov/resource/plmp.m019/?sp=243&amp;r=-0.967,-0.101,2.933,1.818,0"><em>Khamseh, Niẓāmī Ganjavī</em></a><em> from the Library of Congress</em></figcaption></figure><p><strong>✏</strong><a href="https://blogs.nvidia.com/blog/2019/01/24/deep-learning-deciphers-historical-documents/"><strong>Preserving historical manuscripts and the art of calligraphy with AI</strong></a><br>“The Library of Congress has made <a href="https://www.loc.gov/item/prn-19-036/?loclr=ealn">155 rare Persian</a> manuscripts, lithographs, and books dating back to the 13th century available online for the first time. Check out the beautiful passages in Persian, Turkish, and Arabic “written in fine calligraphic styles, and illustrated with miniatures.” While browsing these passages, it occurred to me how difficult it is for me to read different Persian calligraphies. I guess I have to blame it on the force of technology, but can the same force preserve our connection to our old manuscripts and cultural heritage? Here are some of the recent advances in using machine learning for preserving and understanding old manuscripts: <a href="https://wiki.epfl.ch/bigdata2015-deep-learning/documents/project_proposal.pdf">“Word Spotting” in Historical Manuscripts</a>, <a href="https://ieeexplore.ieee.org/document/8389104">Image enhancement of old manuscripts using machine learning</a>, <a href="https://towardsdatascience.com/build-a-handwritten-text-recognition-system-using-tensorflow-2326a3487cd5">Handwritten Text Recognition System</a>. For all you out there: can you think of doing anything with all these beautiful Persian manuscripts? All pages are public domain :)</p><p><em>This has been an excerpt from the Humane AI newsletter, run by me, </em><a href="https://www.royapakzad.co/"><em>Roya Pakzad</em></a><em>. You can subscribe to the newsletter </em><a href="https://royapakzad.us17.list-manage.com/subscribe?u=9138308bb26620c53a0881c20&amp;id=8152ed9f0c"><strong><em>here</em></strong></a><strong><em>.</em></strong></p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=b249274370ba" width="1" height="1" alt=""><hr><p><a href="https://medium.com/humane-ai/when-ai-meets-nowruz-b249274370ba">When AI Meets Nowruz</a> was originally published in <a href="https://medium.com/humane-ai">Humane AI</a> on Medium, where people are continuing the conversation by highlighting and responding to this story.</p>]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[The origin of the Mechanical Turk and Stalin’s “deepfakes”]]></title>
            <link>https://medium.com/humane-ai/humane-ai-newsletter-13-the-origin-of-the-mechanical-turk-and-stalins-deepfakes-59438df68496?source=rss----7ad53ef01cf3---4</link>
            <guid isPermaLink="false">https://medium.com/p/59438df68496</guid>
            <category><![CDATA[human-rights]]></category>
            <category><![CDATA[artificial-intelligence]]></category>
            <category><![CDATA[humaneai]]></category>
            <category><![CDATA[newsletter]]></category>
            <category><![CDATA[machine-learning]]></category>
            <dc:creator><![CDATA[Roya Pakzad]]></dc:creator>
            <pubDate>Wed, 21 Nov 2018 21:09:25 GMT</pubDate>
            <atom:updated>2019-08-22T01:36:10.671Z</atom:updated>
            <content:encoded><![CDATA[<h4>From the 13th edition of my newsletter, <a href="https://mailchi.mp/ce30aaadc443/humane-ai-newsletter-14">Humane AI</a></h4><p><strong>✏ </strong><a href="https://www.theatlantic.com/business/archive/2018/01/amazon-mechanical-turk/551192/"><strong>The Mechanical Turk: “Artificial Artifical Intelligence”</strong></a><br>In the late 18th century, a Hungarian inventor presented a new invention to the Empress of Austria. It was an “intelligent” machine that came to be known as the Mechanical Turk, capable of playing chess at a high level. Over the years, the automaton would defeat opponents ranging from Napoleon to Benjamin Franklin. It was not until decades later that the secret was revealed: a <em>human</em> operator was hidden inside this supposedly thinking machine.</p><p>Two centuries later, are we continuing to hide the role (and, more, importantly the rights) of humans behind our “magical” intelligent machines? For a day’s worth of work on labeling training data, transcribing audio files, and other menial tech tasks, how much do Amazon’s Mechanical Turkers make? Do they have health insurance? Does the US Fair Labor Standards Act apply to them? In <a href="https://www.bbc.com/news/technology-46055595">this piece</a>, the BBC’s David Lee reports on the Kenyans who work annotating images to be used for training self-driving cars. <a href="https://www.theatlantic.com/business/archive/2018/01/amazon-mechanical-turk/551192/"><em>The Atlantic</em></a> also reported on Amazon Mechanical Turkers’ work situations: “3.8 million tasks on Mechanical Turk, performed by 2,676 workers, found that those workers earned a median hourly wage of about $2 an hour. Only 4 percent of workers earned more than $7.25 an hour.”</p><blockquote>If you are interested in learning more about advocating for crowd workers rights, check out <a href="https://www.reddit.com/r/turkernation">Turker Nation</a> and <a href="https://hci.stanford.edu/publications/2015/dynamo/DynamoCHI2015.pdf">We Are Dynamo: Overcoming Stalling and Friction in Collective Action for Crowd Workers</a>.</blockquote><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/0*_KdQHwkxwBNjliW_.jpg" /><figcaption><em>An 1826 print of the Mechanical Turk showing its inner workings (</em><a href="https://www.rc.umd.edu/gallery/chess-player"><em>source</em></a><em>)</em></figcaption></figure><p><strong>✏</strong><a href="https://cdt.org/insight/mixed-messages-the-limits-of-automated-social-media-content-analysis/"><strong>Limitation of Computational Fact Checking and Social Media Analysis</strong></a><br>Recently I learned about a fact-checking tool called <a href="https://www.fakerfact.org/">FakerFact</a>. Using Natural Language Processing, based on training on millions of documents, the tool lets you know what kind of article you are reading: is the article written in a way that focuses on facts, or is it written with a manipulative agenda? FakerFact is not the only one: <a href="http://chequeado.com/automatizacion">Chequeabot</a> in Argentina, <a href="https://fullfact.org/">Full Fact</a> in the UK, <a href="https://osome.iuni.iu.edu/tools/">Botometer</a>, and many more.</p><p>Despite the importance of these tools, perhaps the next step is to understand the limitations of these technologies: who sets the standards of whether an article is truthful or not? Who is the targeted audience: people with a low digital literacy level, or a high one? Can the tool take the context of certain news or certain platforms into account? Is there a chance that the tool disproportionately censors voices from marginalized groups? <a href="https://cdt.org/insight/mixed-messages-the-limits-of-automated-social-media-content-analysis/">This great report</a> by the Center for Media and Technology looks into some of the limitations of content moderation tools.</p><p><strong>✏ </strong><a href="https://www.newyorker.com/magazine/2018/11/12/in-the-age-of-ai-is-seeing-still-believing"><strong>Comrade Stalin, Doctored Images, and Deepfakes</strong></a></p><blockquote>Twelve voices were shouting in anger, and they were all alike. No question, now, what had happened to the faces of the pigs. The creatures outside looked from pig to man, and from man to pig, and from pig to man again; but already it was impossible to say which was which.</blockquote><p>These are the very last lines of <a href="https://en.wikipedia.org/wiki/Animal_Farm"><em>Animal Farm</em></a> by George Orwell. I was 11 when I read it in Farsi; too naive to understand the correspondence between the pig Neopolean in the book and Joseph Stalin of the Soviet Union. But those final lines stayed with me until today. This summer, I visited Stalin’s childhood home and <a href="https://en.wikipedia.org/wiki/Joseph_Stalin_Museum,_Gori">museum</a> in Gori, Georgia. While looking at the rows and rows of photos, I thought about the USSR’s censorship and photo manipulation. Were these strategies ultimately successful in changing people’s mind? In hiding reality? Are there any lessons in history that we can apply to today’s concerns around deepfakes?</p><p>In addition to <a href="https://www.newyorker.com/magazine/2018/11/12/in-the-age-of-ai-is-seeing-still-believing">the New Yorker article</a>, this <a href="https://blog.witness.org/2018/07/deepfakes/">great report by WITNESS</a> gives an overview of deepfakes, the technology behind their creation, and a series of recommendations for tech companies, media literacy educators, and policy-makers.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/500/0*6AS1oCZ9hWOMYgnA.jpg" /><figcaption><em>Stalin airbrushes out his enemies … and he was not the only one! (</em><a href="https://twistedsifter.com/2012/02/famously-doctored-photographs/"><em>Source</em></a><em>)</em></figcaption></figure><p><strong>✏ </strong><a href="http://datanutrition.media.mit.edu/"><strong>A “Nutrition Label” for Datasets</strong></a><br>Recently there have been some excellent discussions among machine learning researchers about improving the transparency and quality of the training data that are fed into machine learning models. MIT’s dataset “Nutrition Label” aims to “create a standard for interrogating datasets that will ultimately drive the creation of better, more inclusive algorithms.” <a href="https://arxiv.org/abs/1803.09010">Datasheets for Datasets</a> and <a href="https://arxiv.org/pdf/1810.03993.pdf">Model Cards for Model Reporting</a> are two other great standardization approaches to “identify how a dataset was created, and what characteristics, motivations, and potential skews it represents.”</p><p><strong>✏ </strong><a href="https://www.theatlantic.com/magazine/archive/2018/04/the-refugee-detectives/554090/"><strong>The Refugee Detectives</strong></a><br>In this piece, <em>Atlantic</em> reporter Graeme Wood shows how Germany’s Federal Office for Migration and Refugees has been using facial recognition tools and language identification as a “lie detector” to evaluate asylum seekers’ applications. From a human rights perspective, there are many concerns around the “black box” nature of these tools, error rates for different languages and dialects, the level of human oversight, and privacy concerns. Germany is not the only country currently deploying AI tools to control its border. Canada, UK, and Australia have also begun using similar technologies. “<a href="https://ihrp.law.utoronto.ca/sites/default/files/media/IHRP-Automated-Systems-Report-Web.pdf">Bots at the Gate: A Human Rights Analysis of Automated Decision Making in Canada’s Immigration and Refugee System</a>” shows how “these initiatives may place highly vulnerable individuals at risk of being subjected to unjust and unlawful processes.”</p><p><strong>✏ </strong><a href="https://5harad.com/papers/fair-ml.pdf"><strong>The Measure and Mismeasure of Fairness: A Critical Review of Fair Machine Learning</strong></a><br>In this new working paper, Stanford researchers Sam Corbett-Davies and Sharad Goel argue that the three common approaches toward ML fairness, including anti-classification (i.e. protected features, like race, gender, and their proxies, are not explicitly used to make decisions); classification parity (equal error rates across protected groups) and calibration (conditional on risk estimates, outcomes are independent of protected features) have significant statistical limitations. “Requiring anti classification or classification parity can, perversely, harm the very groups they were designed to protect; and calibration, though generally desirable, provides a little guarantee that decisions are equitable.”</p><p><em>This has been an excerpt from the Humane AI newsletter, run by me, </em><a href="https://www.royapakzad.co/"><em>Roya Pakzad</em></a><em>. You can subscribe to the newsletter </em><a href="https://royapakzad.us17.list-manage.com/subscribe?u=9138308bb26620c53a0881c20&amp;id=8152ed9f0c"><strong><em>here</em></strong></a><strong><em>.</em></strong></p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=59438df68496" width="1" height="1" alt=""><hr><p><a href="https://medium.com/humane-ai/humane-ai-newsletter-13-the-origin-of-the-mechanical-turk-and-stalins-deepfakes-59438df68496">The origin of the Mechanical Turk and Stalin’s “deepfakes”</a> was originally published in <a href="https://medium.com/humane-ai">Humane AI</a> on Medium, where people are continuing the conversation by highlighting and responding to this story.</p>]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Ethics in Machine Learning]]></title>
            <link>https://medium.com/humane-ai/ethics-in-machine-learning-54a71a75875c?source=rss----7ad53ef01cf3---4</link>
            <guid isPermaLink="false">https://medium.com/p/54a71a75875c</guid>
            <category><![CDATA[artificial-intelligence]]></category>
            <category><![CDATA[machine-learning]]></category>
            <category><![CDATA[ethics]]></category>
            <category><![CDATA[tech-policy]]></category>
            <dc:creator><![CDATA[Roya Pakzad]]></dc:creator>
            <pubDate>Sat, 02 Jun 2018 01:18:11 GMT</pubDate>
            <atom:updated>2018-01-22T07:25:49.431Z</atom:updated>
            <content:encoded><![CDATA[<h4>Interview with Dr. Hanie Sedghi, Research Scientist, Google Brain</h4><p>On a not-very-sunny day in our Golden State of California, I sat down (virtually) with Dr. Hanie Sedghi to discuss the topic of ethics in Machine Learning. Born and raised in Iran, Hanie is a research scientist at <a href="https://research.google.com/teams/brain/">Google Brain</a>, based in Mountain View. Prior to joining Google Brain, Hanie worked at <a href="http://allenai.org/">Allen Institute for Artificial Intelligence</a> in Seattle as a research scientist for two years. She earned her PhD from University of Southern California in Spring 2015. Her dissertation topic is <a href="http://digitallibrary.usc.edu/cdm/ref/collection/p15799coll3/id/610840">Stochastic Optimization in High Dimensions</a>.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*gDE3N_jwVcUzrXtUsZQF7Q.jpeg" /><figcaption>Dr. Hanie Sedghi, Research Scientist, Google Brain (Credit: Hanie Sedghi)</figcaption></figure><p>Hanie moved to the US in 2010 after earning B.Sc. and M.Sc. degrees in Electrical Engineering from Iran’s most prestigious engineering school, Sharif University of Technology.</p><p>After introducing me to her cute Persian cat, Misha— she actually brought him from Iran — we jumped right in. As a side note, Hanie and I talked in our native language, Farsi. Below I have transcribed and translated our conversation, which Hanie reviewed for accuracy.</p><p><strong>Roya Pakzad: Hanie, tell me in your own words, what is fairness in machine learning?</strong></p><p><strong>Hanie Sedghi:</strong> There are many examples and definitions. But let me start with one very well-known example: algorithmic bias in criminal justice systems. In this situation, if we disproportionately feed data about crimes committed by African Americans into a crime prediction model, then of course our model’s prediction will be biased against black communities. This is a skewed sample. On the other hand, if minority groups are underrepresented in our data samples, for example women or transgender individuals, then again, we can expect that our model will have a bias.</p><p><strong>Roya: Are there any mathematical solutions for addressing that?</strong></p><p><strong>Hanie:</strong> To some extent yes. There are three different approaches:</p><p>1) Post processing in terms of calibration of our model. What this means is that, we calibrate classifiers parameters such that it has the same acceptance ratio for all subgroups of sensitive features, e.g. race, sex, etc.</p><p>2) Data resampling to remove skewed sample. But, for many reasons, collecting more data is not very easy and sometimes causes problems for individuals.</p><p>3) Causal reasoning: We capture different paths in a causal graph that can lead to the same observational data. This basically means to model possible factors such that sex, race and other sensitive features to make sure their impact is captured and does not directly affect the result variable.</p><p><strong>Roya: What are some of the challenges in modeling fairness in decision-making algorithms?</strong></p><p><strong>Hanie:</strong> One very important issue is the lack of a concrete definition of “fairness.” I can tell you that there are a number of definitions and sometimes research groups are not on the same page when it comes to the definition of fairness. When you don’t have a clear definition, then how can you model it correctly?</p><p>The other issue is the need for collaboration between social scientists and AI researchers. You know, you can’t expect AI researchers themselves to come up with a clear understanding of fairness. Not only we need people in social sciences to collaborate with us in defining these words, but also we need to keep this collaboration all along to the end of the product research and development.</p><blockquote>“One very important issue is the lack of a concrete definition of fairness.”</blockquote><p>But it’s important to note that some collaborations between AI researchers and social scientists are already underway. For example, Solon Barocas (Cornell University) and Moritz Hardt at UC Berkeley have been working on the issue of defining and modeling fairness in active collaboration with social scientists.</p><p><strong>Roya: But they are academics. They are not expected to meet corporate production deadlines or offer super-practical and financially-feasible solutions. They might not be under pressure from the AI competition that we are witnessing in the big tech companies. How do you think we can encourage data scientists and engineers to be aware of these issues who might be under tough deadline pressures?</strong></p><p><strong>Hanie:</strong> Well, there should definitely be some sort of training on ethics and fairness for them.</p><p><strong>Roya: How to make sure those trainings are effective? I, myself went through some types of company’s mandatory online 45 minute quizzes on ethics in the workplace. I confess I clicked all answers until I get the right one, without being actually trained!</strong></p><p><strong>Hanie:</strong> I understand. This is a question that I can’t answer but I also would really like to know how to achieve this. But I do know that, with respect to bias and fairness, as long as our training is in the form of someone lecturing about the basics of gender or racial bias in society, that training is not likely to be effective. When you warn us about the issues without giving any solutions or approaches to resolve it, then how can that training be fully effective?</p><p>I also think first we should have a practical guideline. Currently, there is not any comprehensive guideline (that is actually implementable) in the field of ethics in machine learning. Let me give you an example: differential privacy. It means making sure that the data you have do not reveal the identity of the persons to whom the data belongs. It has a clear mathematical definition. So, for example, once for the Netflix Prize competition, to build a recommender system, we, as researchers, were able to identify that their dataset violates differential privacy. But for fairness, we currently do not have this.Consequently, it cannot be easily taught and monitored, and audited.</p><p><strong>Roya: You are an Iranian and you finished your B.Sc. and M.Sc. in Electrical Engineering. You also worked in the industry. How do you see the state of AI research in Iran?</strong></p><p><strong>Hanie:</strong> Well, I was in Iran recently and had a chance to talk to some AI researchers there. There are great start-ups working mostly on developing apps similar to everything we use here. But regarding AI research, the progress is behind. Machine Learning itself is still relatively a new field. One reason for the gap could be the US sanctions against access to certain research papers, and, more importantly, to cloud computing platforms.</p><p><strong>Roya: I’m sure you know better than me that there is a lack of representation of women in tech. There is also a high rate of women leaving the tech industry. How can we address both issues?</strong></p><p><strong>Hanie:</strong> Regarding lack of representation, I think we should motivate girls from earlier ages in the STEM fields. There are great initiatives by Google and Twitter currently underway. For example, they invite young girls to their campuses and give them small projects that motivate them to think about machine learning as a field that is capable of solving many real world problems.</p><p>Having strong role models is important too. Sometimes this lack of representation leads to lack of self confidence in women. They might feel frustrated about not being heard and intellectually respected. Eventually, they might leave the field. So, it is very important to have strong female role models to look up too. One of my own childhood role models was Marie Curie. In ML also, I had great role models such as Anima Anandkumar and Jennifer Chayes. I learned a lot from their strong personalities and self-confidence.</p><p><strong>Roya: A final question: are you concerned about the future of AI?</strong></p><p><strong>Hanie:</strong> I don’t believe in picturing AI as a devil. We are the teacher and we have control over what we teach.<strong> </strong>I believe that Machine Learning has great potential to make our life easier, from simple tasks such as weather forecasting and AI assistants to self driving cars. But I do think that we still have a lot of work to make AI intelligence reach anything approaching the level of humans. For example, you can have your AI answer your questions but you cannot yet expect it to fully reason. There are still many many difficult unanswered challenges and that’s what making this field the most interesting to me!</p><p>We wrapped up here. This is the first conversation of my interview series for my newsletter <a href="https://www.royapakzad.co/newsletter/">Humane AI</a>. I will continue talking with both policy and technical experts in the field of ethics of AI in future installments. Tune in to know their opinions about many issues including cybersecurity and AI, private sector approaches in addressing such ethical challenges, Human rights and AI for social good, and much more. To subscribe to the newsletter, click <a href="https://royapakzad.us17.list-manage.com/subscribe?u=9138308bb26620c53a0881c20&amp;id=8152ed9f0c">here</a>.</p><p>You can find Dr. Hanie Sedghi on <a href="https://twitter.com/HanieSedghi?lang=en">Twitter</a> and <a href="https://www.google.com/url?sa=t&amp;rct=j&amp;q=&amp;esrc=s&amp;source=web&amp;cd=2&amp;cad=rja&amp;uact=8&amp;ved=0ahUKEwjylvP-sePYAhUV6WMKHQCEC08QFggzMAE&amp;url=https%3A%2F%2Fwww.linkedin.com%2Fin%2Fhanie-sedghi-71bb2582&amp;usg=AOvVaw0YEx3cTOS3g83S7Bs38Baa">LinkedIn</a>.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=54a71a75875c" width="1" height="1" alt=""><hr><p><a href="https://medium.com/humane-ai/ethics-in-machine-learning-54a71a75875c">Ethics in Machine Learning</a> was originally published in <a href="https://medium.com/humane-ai">Humane AI</a> on Medium, where people are continuing the conversation by highlighting and responding to this story.</p>]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Artificial Intelligence and Corporate Social Responsibility]]></title>
            <link>https://medium.com/humane-ai/artificial-intelligence-and-business-social-responsibility-69d6299b4d9d?source=rss----7ad53ef01cf3---4</link>
            <guid isPermaLink="false">https://medium.com/p/69d6299b4d9d</guid>
            <category><![CDATA[business-and-human-rights]]></category>
            <category><![CDATA[corporate-sustainability]]></category>
            <category><![CDATA[artificial-intelligence]]></category>
            <category><![CDATA[ethics]]></category>
            <dc:creator><![CDATA[Roya Pakzad]]></dc:creator>
            <pubDate>Sat, 02 Jun 2018 01:17:56 GMT</pubDate>
            <atom:updated>2023-06-11T05:56:38.866Z</atom:updated>
            <content:encoded><![CDATA[<h4>Interview with Dunstan Allison-Hope, Managing Director, BSR</h4><p>Accountability.</p><p>How many times have you heard this word when talking about a company’s role in our society? Either during a casual talk with a friend about self-driving cars, or reporting on automation and the future of work, or in corporate boardrooms, or on the streets protesting for your privacy protection, I bet you have heard and used this sentence: “Companies should be held accountable.”</p><p>The rapid advancement in AI technologies has opened up new challenges for companies with regards to their social responsibilities. A company whose motto once was to “move fast and break things” now finds itself compelled by their customers, civil society groups, governments, shareholders, and perhaps their own conscience to hit the brakes, look backward, and move cautiously toward future.</p><p><a href="https://www.bsr.org/en/about/staff-bio/dunstan-allison-hope">Dunstan Allison-Hope</a> and the organization he’s a part of, <a href="https://www.bsr.org/en/">BSR</a> (Business for Social Responsibility), is one of those forces working to direct companies towards a better path in terms of their commitment to ethics, human rights, and sustainability. As a managing director at BSR, Dunstan has worked on diverse range of corporate social responsibility issues including privacy and freedom of expression, human rights, stakeholder engagement, and transparency reporting in different parts of the world.</p><p>Below is my conversation with Dunstan about tech companies’ social responsibility with regards to AI and other emerging technologies.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/343/1*MbRz59EtNWJ3HndFWWZLBQ.jpeg" /><figcaption>Dunstan Allison-Hope, Managing Director, BSR (Credit: BSR)</figcaption></figure><p><strong>Roya Pakzad: Dunstan, I have been following your work for the past year and based on your publications, I realized you have a strong interest in corporate social responsibility with regards to Artificial Intelligence. Why is that?</strong></p><p><strong>Dunstan Allison-Hope</strong>: I’ve worked a lot with tech companies on privacy and freedom of expression, and AI is a natural extension of that. These big technologies are going to change businesses and raise new social, ethical, environmental, and sustainability issues and that is what we at BSR care about.</p><p><strong>Roya: What are some examples of those risks that companies should care about?</strong></p><p><strong>Dunstan</strong>: Specific risks vary company to company and industry to industry, but it’s fair to say that the product that is launched might have adverse human rights impacts. These could include privacy violations, freedom of expression issues, discriminatory impacts, and impacts on children’s rights online. Children increasingly participate online and are active digitally. They have rights and they are an especially vulnerable group.</p><p><strong>Roya: </strong><a href="https://business-humanrights.org/en/un-guiding-principles"><strong>The UN Guiding Principles on Business and Human Rights</strong></a><strong> (UNGP) has defined certain principles in terms of corporate responsibility to <em>respect</em> human rights. How can those principles be applied here?</strong></p><p><strong>Dunstan</strong>: UNGPs talk about the due diligence process — that means having a commitment to human rights and assessing actual and potential adverse human rights impacts. Companies should engage to identify those potential adverse impacts and put in place mitigation plans to address them. UNGP is pretty clear on that, but I think technology is complex and this makes it difficult in practice. The challenge for technology companies is that a lot of impacts happen through the products’ use phase. So, how can you assess the potential adverse human rights impacts when you don’t know how these products are going to be used? That’s challenging, but it should not be an excuse for companies. Companies can put policies and processes in place to prevent those risks and make sure all those [adverse impacts] are factored in during product design and release.</p><p><strong>Roya: I understand that those policies and processes should be based on certain principles and guidelines. Currently there are several guidelines, code of ethics, and principles. But, I don’t quite see a concrete implementation plan of those guidelines on industry level. How can companies implement those guidelines into their practices?</strong></p><p><strong>Dunstan</strong>: The current principles are very high level but there is enough similarities between them at this stage that they set very good direction for companies. I think the challenge is the fact that we don’t know what “good” looks like in terms of how to actually implement them. For example in labor standards and supply chain, we already know how companies should implement them. When it comes to AI, it’s so new. What we need is the combinations of real life examples and case studies. By looking at different use cases and real life examples you might realize some principles need to change in practice.</p><p>We will also need industry specific version of these guidelines. For example, how to apply good ethical AI guideline for financial services, how to apply it for criminal justice, how to apply it to the context of social media platforms. For example, in the US, as a result of civil rights protections, there are various things that companies are not allowed to do and AI is subject to those rules. But they might have some loopholes and risks because those principles are written for a different age and government tends to be behind when it comes to technological developments. For example, have a look at the Net Neutrality debate, telecom regulations, online privacy rules. Government tends to move more slowly than technology does.</p><p><strong>Roya: In the past you proposed the concept of </strong><a href="https://www.bsr.org/our-insights/blog-view/human-rights-by-design"><strong>Human Rights by Design</strong></a><strong> for technology companies. How can companies apply that concept in their social responsibility efforts?</strong></p><p><strong>Dunstan</strong>: So in normal human rights impact assessments, it is typical for a company to take a cross functional approach. It might be run by the legal team or public affairs group, but typically human rights impact assessment is overseen by cross functional teams. The the human resource team, legal, public affairs, social responsibility, and supply chain groups all typically participate. But engineering function or product development teams are usually absent. This is the blind spot. You might not need to change the actual impact assessment tools very much, you might not need to change the questions very much, but you should change who is participating. And I’m not convinced that is happening. There should be different sets of communities that should get involved, including engineers, data scientists, and product development teams in general.</p><p>The other issue is that in practice a lot of human rights impact assessments are on the market or country or company overall. They are rarely on specific products or product categories. I think we need more human rights impact assessment at the product level. For example, on new types of communication products, and new types of big data and analytic tools that companies didn’t have before. Products themselves should be subject to assessments — perhaps an extended version of today’s privacy by design methods.</p><p><strong>Roya: Any successful examples among companies?</strong></p><p><strong>Dunstan</strong>: Some, in the context of broader projects, but not nearly as directly as would be ideal. Microsoft is doing Human Rights Impact assessment for AI which will be very interesting to see what they’ll conclude. That’s a good example.</p><p><strong>Roya: With regards to applying human rights standards, do you think technology companies respond better to voluntary regulations or hard regulations?</strong></p><p><strong>Dunstan: </strong>I think both is the answer! I read a very interesting article the other day — I believe I linked to it via your newsletter — about regulating specific topics, such as access to credit, rather than AI overall<strong>, </strong>which might cause many different types of unintended consequences. I also think that whether voluntary or mandatory, approaches need to work with the grain of existing internationally agreed frameworks for sustainable business, such as the UN Guiding Principles, the OECD Guidelines for Multinational Enterprises, and the G20/OECD Principles of Corporate Governance. Personally, I’m a big fan of disclosure requirements and transparency as drivers of improved performance and accountability.</p><p><strong>Roya: Any final thoughts to share?</strong></p><p><strong>Dunstan: </strong>There is a need to bring together more actors more deliberately than what is currently happening. Sustainability teams and social responsibility teams have long history of engaging with big social challenges and they need to be more engaged in the ethics of AI. But that debate also needs engineers and data scientists. These kinds of multi-disciplinary approaches are essential and there is room for improvement there.</p><p>We wrapped up here. This conversation was part of the interview series for my newsletter <a href="https://royapakzad.substack.com/">Humane AI</a>. I will continue talking with both policy and technical experts in the field of ethics of AI in future installments. Tune in to know their opinions about many issues including cybersecurity and AI, machine learning in disaster management and humanitarian context, Human rights and AI for social good, and much more. To subscribe to the newsletter, click <a href="https://royapakzad.substack.com/">here</a>.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=69d6299b4d9d" width="1" height="1" alt=""><hr><p><a href="https://medium.com/humane-ai/artificial-intelligence-and-business-social-responsibility-69d6299b4d9d">Artificial Intelligence and Corporate Social Responsibility</a> was originally published in <a href="https://medium.com/humane-ai">Humane AI</a> on Medium, where people are continuing the conversation by highlighting and responding to this story.</p>]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Human Rights Advocacy in the Dawning Age of AI]]></title>
            <link>https://medium.com/humane-ai/human-rights-advocacy-in-the-dawning-age-of-ai-83274c922d41?source=rss----7ad53ef01cf3---4</link>
            <guid isPermaLink="false">https://medium.com/p/83274c922d41</guid>
            <category><![CDATA[human-rights]]></category>
            <category><![CDATA[humane-tech]]></category>
            <category><![CDATA[data-analysis]]></category>
            <dc:creator><![CDATA[Roya Pakzad]]></dc:creator>
            <pubDate>Sat, 02 Jun 2018 01:17:44 GMT</pubDate>
            <atom:updated>2018-02-16T22:35:49.517Z</atom:updated>
            <content:encoded><![CDATA[<h4>Interview with Dr. Alexa Koenig, Executive Director, UC Berkeley Human Rights Center</h4><p>On a day when blooming fruit trees, singing robins, and bright yellow flowers of mimosa trees teamed up to bring an untimely spring to Berkeley, I sat down with <a href="https://www.law.berkeley.edu/research/human-rights-center/about-hrc/faculty-staff/alexa-koenig/">Dr. Alexa Koenig</a> to have a very timely conversation on the intersection of emerging technologies and human rights.</p><p>Dr. Alexa Koenig is the Executive Director of the <a href="https://www.law.berkeley.edu/research/human-rights-center/">UC Berkeley Human Rights Center</a>. Her journey as a human rights scholar began in the aftermath of the 9/11 attack. As a PhD student, Alexa was offered a position at the Human Rights Center at Berkeley to work on War on Terror related investigations. In the meantime, she started working with <a href="https://www.usfca.edu/law/faculty/peter-honigsberg">Peter Jan Honigsberg</a> at the University of San Francisco to develop an oral history project, <a href="http://witnesstoguantanamo.com/">Witness to Guantanamo</a>, to interview and record the experience of detainees in Guantanamo Bay.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/640/0*cZPX1pWB_LxEoMP9." /><figcaption>Dr. Alexa Koenig (Credit: Human Rights Center, UC Berkeley)</figcaption></figure><p>Her interest in harnessing new technologies for human rights blossomed by cooperating with the Office of the Prosecutor at the International Criminal Court, as a member of the Technology Advisory Board. She and her team held a number of workshops and invited technologists to understand how to use different techniques such as remote sensing, satellite imagery, collecting video footage and data analytics to document human rights abuses and war crimes.</p><p>Bellow is my conversation with Alexa on her valuable insights and experience in utilizing the power of technology in human rights and humanitarian efforts.</p><p><strong>Roya Pakzad: Alexa, could you tell me what you and your students do at your </strong><a href="https://www.law.berkeley.edu/research/human-rights-center/programs/technology/human-rights-investigations-lab-internships/"><strong>human rights investigations lab</strong></a><strong>?</strong></p><p>Alexa Koenig: [With the advent of social media] there was suddenly a wealth of information coming out in terms of video footage and photographs that were invaluable for investigating human rights abuses and war crimes in conflict zones. In September 2016, Andrea Lampros and I launched the human rights investigation lab as a pilot effort to work with other human rights non-profits. We wanted to see if students can be trained to provide critical capacity for human rights investigation and accountability efforts. We started with Amnesty International, and they began sending us footage of different crises that students could verify through different techniques, such as checking the source of the video, using tools like SunCalc to verify the time of the day when the footage was taken.</p><p>At our lab, students from computer science and engineering backgrounds, anthropologists, law students, and journalists are brought together to combine their know-how about how to do this work and take their research and investigations to that next level.</p><p><strong>Roya: This is one of the successful examples of using technology for human rights. But, I have seen many tech oriented human rights projects that in practice were not very effective and sustainable. How can we address that concern?</strong></p><p>Alexa: There have been so many lessons learned over the past ten years about how to ethically and responsibly use these new technologies. Back in 2012 to 2014, people were trying to design new apps to document what’s happening around the world but were not always sensitive to social context. One example is a company creating an app to document sexual violence but rolling it out in a country where only men had access to cellphones. Or, in the middle of a crisis, it’s very unlikely for someone to download an app and learn how to use it and trust it. So, one thing that I’ve been personally interested in is that how do we get individuals who were not pipelined in human rights practice to be more sensible on ethics and social issues that ideally they will think about when they are designing new technologies.</p><p><strong>Roya: And the role of human rights funders here?</strong></p><p>Alexa: Yes, there has been an enormous diversion of resources from traditional methods of human right practice toward investing in these very sexy, shiny new tools. But the risks of innovation in the human rights space are arguably more acute than when you are talking about innovating day-to-day practice in the context of [places] not being hit by crisis. The biggest message to funders is to find people who know human rights and do it well. Sometime the answer might be no technology at all.</p><p><strong>Roya: Berkeley is positioned close to Silicon Valley. Do you try to have these types of conversations and partnerships with technologists to inform them about these lessons learned?</strong></p><p>Alexa: I think there are still silos in practice across the Bay Area. One effort that we’ve been really experimenting with is a new nonprofit spun out of the UC Berkeley campus called Archer [<a href="https://www.archerimpact.com/">website</a>.] This is a team of about 30 computer science and electrical engineering students and a couple of students from other disciplines. Three students had actually been on the ground in the South of France when a terrorist attack happened and one of the students was killed during that attack. So, the two students who came back were thinking how to tap into this extraordinary talent here and use their degrees to fight against terrorism. Over the past year we had them embedded with our human rights investigations lab. So they are helping our team to understand how to use technology to make our work more efficient and effective.</p><p>One of the issues with working in a Law School is that law is notoriously behind the curve in terms of adopting technology into practice. So we’ve been thinking that this partnership will help lawyers and political science and anthropology students understand the social context and what needs to be accomplished, and computer scientists can [point out] when this is something that could be automated to be more effective. Many of these students may go on to pursue high profile and high power tech careers. We are hoping that the insights they get from working on these sensitive human rights investigations will be able to inform everything they do in the future in their tech careers.</p><p>The other area that we have been working in is thinking through the large quantities of information that’ve been getting shared on Youtube, Facebook, and Twitter. [These services] are extraordinarily valuable for getting justice for human rights abuses. So, the big conversations that have been happening in the human rights space is how can we work directly with corporations to make sure that if information had to be taken down because of violations of companies’ terms of service, how can they be preserved so it can still be used for our investigative purposes? Or, how can human rights communities capture this content at the moment of uploading, such that even if the videos <em>do</em> have to be taken down, they are still preserved and the stories behind them can be told.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/960/1*G1O3wrU0eBch81ph3nT7jA.png" /><figcaption><em>Human Rights Investigations Lab (Credit: </em>California Magazine / Just In)</figcaption></figure><p><strong>Roya: As you mentioned, law is behind in this regard. It makes me think about international human rights law and the role that the Human Rights Council in particular and the UN in general plays here. What do you expect to see from them to cope with these technological changes?</strong></p><p>Alexa: I think there’s been some conversation within the UN and other large bureaucratic organizations about whether there is a need for other new technologies to be developed to improve the process of gathering information for human rights investigations. Many of us in human rights communities feel that the technologies are largely there — they just need to be strengthened and constantly refined. But, I think it’s become very difficult to map the entire network of people who are working on these technologies for human rights. What I would hate to see more than anything is new technologies being developed when they’ve already been developed.</p><p>I think there is a role for groups like the United Nations and other bodies to play here. Because they have a sort of 30,000-foot overview and everybody knows who they are, they could be these umbrella groups to gather all the information on who’s doing what.</p><p><strong>Roya: The field of machine learning has unleashed numerous opportunities for data analysis. What are some of your hopes to further usage of AI/ML in your work?</strong></p><p>Alexa: As we’re moving forward, the more that we see artificial intelligence and machine learning as being an interactive practice between humans and machines, the better it is going to be, as opposed to seeing machines as replacing humans. Our interpretive processes are somewhat quantitative that can be automated, but they are also highly qualitative and subjective. Machines and humans can be checks and balance for each other.</p><p>One thing I would love to see is for machines to be able to detect and flag videos that are extraordinary graphic so there’s a warning on them before human beings have to work with that footage. One of the things that resiliency experts have observed is that when people are trying to work with highly graphic footage, and you’re surprised by the content, it seems to have a much larger negative impact than when they’re told what they’re going to see, such as a beheading or something else. You need to be prepared for that.</p><p>Another thing that I would love to see is a “Google translate for algorithms.” A lot of the biases that we are all reading about that go undetected might be detected by human rights practitioners if they could understand how the systems they are using actually work.</p><p><strong>Roya: What are some of your concerns?</strong></p><p>Alexa: I think my biggest concern right now with the proliferation of AI and machine learning is on how these technologies are potentially shifting our focus. For example, are we going to begin paying attention only to the crises where smartphones proliferate — the ones that are, in some ways, low hanging fruit? Does that mean the communities where there aren’t common use of digital technologies will be less visible or set aside because “they are not easy to deal with?” That’s not only about the location but the kinds of crimes and harms as well. For example sexual violence can be very difficult to document with digital technologies, but crimes like the dropping of chemical weapons might be less difficult to document with video footage. So does this mean that there’s going to be a growing divide as we shift our focus towards these technologies that make harms visible? Does that mean those [crimes like sexual violence] will be even more in the background?</p><p><strong>Roya: Do you have any final message to people in this ecosystem of technology and human rights?</strong></p><p>Alexa: I think my biggest message to people in this field of work is to get out of whatever bubbles they are in — and we’re all in at least one.</p><p>If engineers and computer scientists start working on the ground with human rights practitioners, if lawyers start partnering up with people who understand emerging technologies and start thinking a little bit more for the future, if all of these individuals can better understand the experiences of survivors of human rights abuse, I think we’re going to be able to do incredible things safely and efficiently.</p><p>We wrapped up here. This interview has been lightly edited and condensed.</p><p>This conversation was part of the interview series for my newsletter <a href="https://www.royapakzad.co/newsletter/">Humane AI</a>. I will continue talking with both policy and technical experts in the field of ethics of AI in future installments. Tune in to know their opinions about many issues including cybersecurity and AI, History of technology and its connection to AI, Digital Humanities, and much more. To subscribe to the newsletter, click <a href="https://royapakzad.us17.list-manage.com/subscribe?u=9138308bb26620c53a0881c20&amp;id=8152ed9f0c">here</a>.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=83274c922d41" width="1" height="1" alt=""><hr><p><a href="https://medium.com/humane-ai/human-rights-advocacy-in-the-dawning-age-of-ai-83274c922d41">Human Rights Advocacy in the Dawning Age of AI</a> was originally published in <a href="https://medium.com/humane-ai">Humane AI</a> on Medium, where people are continuing the conversation by highlighting and responding to this story.</p>]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[In the age of growing cyber-threats, how do philanthropic organizations have our back?]]></title>
            <link>https://medium.com/humane-ai/philanthropic-role-in-our-digital-civilization-3e13556a7890?source=rss----7ad53ef01cf3---4</link>
            <guid isPermaLink="false">https://medium.com/p/3e13556a7890</guid>
            <category><![CDATA[cybersecurity]]></category>
            <category><![CDATA[philanthropy]]></category>
            <dc:creator><![CDATA[Roya Pakzad]]></dc:creator>
            <pubDate>Sat, 02 Jun 2018 01:17:28 GMT</pubDate>
            <atom:updated>2018-03-08T07:08:36.505Z</atom:updated>
            <content:encoded><![CDATA[<h4>Interview with Eli Sugarman, Cyber Initiative Program Officer, the Hewlett Foundation</h4><p>In November of last year, the <a href="https://www.cnas.org/research/technology-and-national-security/artificial-intelligence">Center for a New American Security (CNAS)</a> was a gathering point for prominent voices in technology companies, academic institutions, and government agencies who sat down and discussed the implications of the AI revolution on global security. CNAS is far from being the only think-tank in the US that researches issues related to technology policy. Think tanks and academic institutions carry a heavy weight of responsibility for researching the implications of emerging technologies. Financially speaking, many of these groups count on the generosity of philanthropic foundations. The Hewlett Foundation is one of the leading organizations in this field, having maintained its own <a href="https://www.hewlett.org/strategy/cyber/">Cyber Initiative</a> program since 2014.</p><p>I sat down with <a href="https://www.hewlett.org/people/eli-sugarman/">Eli Sugarman</a>, a former foreign affairs officer at the US State Department, whose passion for international security brought him from Washington DC to Stanford University’s Law School, and now to the Hewlett Foundation as the program officer of their Cyber Initiative program.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*-ast2LAF50Y8LuGPT3nm7g.jpeg" /><figcaption>Eli Sugarman, Cyber Initiative Program Officer, the Hewlett Foundation</figcaption></figure><p><strong>Roya Pakzad: What are some of the applications of AI in the context of cybersecurity?</strong></p><p><strong>Eli Sugarman</strong>: I think artificial intelligence and automation are going to change the world in many ways. When it comes to cybersecurity, one of the real challenges is that there are so many ways that your organizations can be compromised, that somebody can get into your networks and steal your data, surveil, or commit other malicious activities. So what you need to do is to detect them and prevent them from accessing the sensitive information.</p><p>When automation and AI comes [into the picture], it’s really hard if you just rely on people to do all that, because you can’t. When you have a complex organization with hundreds of thousands of endpoints and devices, you have to find an automated way. A specific example is the US Defense Advanced Research Projects Agency’s [DARPA] competition called the DARPA Grand Challenge. What they did is a “Capture the Flag” competition where you had to program your AI bot to attack a network, capture something, and exfiltrate it, while also defending and patching your own system from attacks at the same time. It was a purely automated offensive and defensive capture the flag.</p><p>So, I think one thing we will see is more and more of the basic and mid-tier offense and defense automated. And I think that is going to be a huge sea change which leads to all sorts of unintended consequences and escalatory dynamics [between machines’ attacking each other] that is hard to predict. So, that’s one example that I think how automation is going to change cybersecurity.</p><p><strong>Roya: I’m glad you brought up DARPA, because it shaped a question in my mind about the role of US governmental agencies regarding emerging technologies. Thinking about certain policies and data de-regulation, for example with respect to Net Neutrality and data encryption, I can see a partisan pattern in US policymakers viewpoints on emerging technologies which is sometimes unproductive. The Hewlett Foundation funds some large think-tanks in Washington DC. For your Cyber Initiative program, how do you try to make sure your funding will lead to productive conversations and non-partisan policymaking?</strong></p><p><strong>Eli</strong>: There are a couple of points. One is that we are certainly non-partisan. The second thing is we try to find the smartest people in the best universities and best think tanks from different views and fund both of them at the same time to offer different approaches into the policy conversation. So if you fund different perspectives and purposefully do that, you know in theory you get a better outcome because each of those views have to take into account the other and explain why one is better.</p><p>The third thing is [cyber policy] issues are not as partisan as taking a “Democrat position” or “Republican position.” You saw that recently with the debate in the US Congress about a certain surveillance law where you had a mix of Democrats and Republicans supporting and a mix of them opposing the specific intelligence program called FISA section 702 [<a href="https://intelligence.house.gov/fisa-702/">link</a>].</p><p>So, in my mind there is a lot of education that still needs to happen to understand these issues. But how do you work through those tensions and trade off to find the right policy solutions? We honestly don’t know, but by being very open and transparent and funding different points disagreeing with each other, we are trying to engage in education.</p><p><strong>Roya: When you look at the rapid advancement of technologies, it is obvious that governments and policymakers are behind in coming up with timely policies. You mentioned that there is still lots of education that needs to be done. How does your program try to address this?</strong></p><p><strong>Eli: </strong>What we are trying to do is to find the resources and tools that policymakers need. So we’ve done surveys to ask policymakers what education would be helpful and tried to support the creation of that technical assistance. Some examples: we fund Stanford University and the Hoover Institution, a non-partisan think tank, to do what they call boot camps for congressional staff. They come for four hours for three days. The staff members receive non-partisan tailored educational sessions on cybersecurity topics. Likewise, we fund the Woodrow Wilson Center in Washington, DC to do something similar. So, for example if you want to know about encryption? Here is four hours with the former deputy director of the NSA who explains how encryption works, why it matters for national security, why it matters for businesses, and the complexity of the issue, so that when you go back to your congressional office you then have some knowledge, you have been given some resources that you can read and share with other people and also you know some experts to call to help you with your jobs. We found these bootcamps and training sessions really effective.</p><p>We also found there is a new class of written products that need to be delivered. If you are a busy senior government official you don’t have time to read a full book. We’ve funded an interesting project at the Carnegie Endowment for International Peace called cyber analogies [<a href="http://carnegieendowment.org/2017/10/16/introduction-to-understanding-cyber-conflict-14-analogies-pub-73392">Introduction to Understanding Cyber Conflict: 14 Analogies</a>] where it gives you a 10–15 page chapter that says cybersecurity is similar to this other thing you know very well or different that helps you to understand it. So how is cybersecurity similar or different than drones? Or counter terrorism? Our grantees are looking to find creative ways to tell stories and educate in ways beyond saying here is a 100 page report and go read it.</p><p><strong>Roya: As you know better than me, the effectiveness of policies and regulations are heavily dependent on constant measurement and evaluation. The field of cybersecurity is very new and we are seeing rapid advancement in it because of proliferation of AI applications. But in some ways these changes and their impacts need to be measured. What can foundations do in providing fair, comprehensive, and non-partisan measurement and evaluation that leads to better policy making decisions?</strong></p><p><strong>Eli: </strong>I think that foundations can do a lot to promote data driven and empirical research. An example: another problem in cybersecurity is that you want your security researchers to want to do research and find vulnerabilities of a system. [But] when they do, companies threaten them with legal action and say you have broken the law by de-compiling my code and finding this vulnerability. People talk about that being a problem, but nobody ever tried to quantify it. So, we funded a project at UC San Diego to basically try to quantify how much legal risk security researchers face, which then helps answer the question of what we need to do about it.</p><p>So oftentimes we assume we have problems, but they are not actually being proven based on data. That’s a huge problem when it comes to cybersecurity issues. If you want to talk about how to measure how many cyber conflicts happened, there are not even agreements about what people are measuring or who is tracking what.</p><p>If you look at data around cyber attacks, it’s really uneven. I think the whole lack of data and empirical research is a big problem. The real question is, at the end of the day, do the decision makers actually want that? Do they care? Sometimes yes, sometimes no, and that’s a whole other conversation.</p><p><strong>Roya: When I think about cybersecurity initiatives, I still feel that people are not fully engaged in conversations and decision making process. How can we improve their awareness and engagement?</strong></p><p><strong>Eli: </strong>Yes, the engagement is really uneven. So one of the big problems is that there is no real effective global governance system for the Internet or cybersecurity. You have ICANN, you have certain standards and bodies to manage technical aspects, but in terms of policy and real rules to keep people safe, there is no real global governance system. The answer is that it’s not really clear who decides on what, when. You have government, you have the United Nations based on what member states want. It’s a huge mess.</p><p>Generally speaking, Internet users are not fully aware, are not fully engaged, and don’t always have a voice.</p><p><strong>Roya: Has the Hewlett Foundation done any projects for public education?</strong></p><p><strong>Eli: </strong>A little bit. But our focus is not much on public engagement. Because to do that well across a broad group of people it costs a lot of money. So we try to spend our resources on experimenting and figuring out new ways of doing some really impactful research. Our focus is on the education of influential elites: Senior government officials, C-suite executives, journalists. Because it’s more manageable based on our size as a grantmaker.</p><p>If you think about millions of dollars spent on public awareness campaigns, for example to educate people that they should wear seat belts or not do drugs, they must have cost tens of million and hundreds of million dollars. I guess the closest we’ve done is funding UCLA to basically do a bootcamp for Hollywood screenwriters and show runners. They brought cyber experts to LA to help educate them so they depict these issues in movies and TV shows more accurately. And that’s [working] a bit indirectly help educate people.</p><p><strong>Roya: What are some of the challenges in your job?</strong></p><p><strong>Eli: </strong>Personally what’s hard is to say no to people. There are so many worthwhile groups and organizations that want to do impactful work and sometimes you just have to say no because we don’t have unlimited money.</p><p>More substantively challenging is that the community of the people who care about cyber issues is pretty big, and also diverse, and also very siloed. Trying to find ways to build trust and bring those different communities and stakeholders is really hard. Because they are operating typically in a separate world with their own vocabularies and their own social networks. It’s really complicated and difficult.</p><p><strong>Roya: And your final message to the readers?</strong></p><p><strong>Eli: </strong>These issues [broadly defined as cyber threats] are obviously not going away. They are going to be more important and more global. It’s great for people to think more and more about these issues and find right way to get involved. Just don’t be passive and feel like you don’t have a voice.</p><p>We wrapped up here. This conversation was part of the interview series for my newsletter <a href="https://www.royapakzad.co/newsletter/">Humane AI</a>. I will continue talking with both policy and technical experts in the field of ethics of AI in future installments. Tune in to know their opinions about many issues including computational disinformation campaigns, history of technology and machine learning, AI for social good, and much more. To subscribe to the newsletter, click <a href="https://royapakzad.us17.list-manage.com/subscribe?u=9138308bb26620c53a0881c20&amp;id=8152ed9f0c">here</a>.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=3e13556a7890" width="1" height="1" alt=""><hr><p><a href="https://medium.com/humane-ai/philanthropic-role-in-our-digital-civilization-3e13556a7890">In the age of growing cyber-threats, how do philanthropic organizations have our back?</a> was originally published in <a href="https://medium.com/humane-ai">Humane AI</a> on Medium, where people are continuing the conversation by highlighting and responding to this story.</p>]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Designing for Ethics]]></title>
            <link>https://medium.com/humane-ai/designing-for-ethics-706efa1a483e?source=rss----7ad53ef01cf3---4</link>
            <guid isPermaLink="false">https://medium.com/p/706efa1a483e</guid>
            <category><![CDATA[social-media]]></category>
            <category><![CDATA[ethics]]></category>
            <category><![CDATA[computational-propaganda]]></category>
            <category><![CDATA[tech]]></category>
            <dc:creator><![CDATA[Roya Pakzad]]></dc:creator>
            <pubDate>Sat, 02 Jun 2018 01:17:07 GMT</pubDate>
            <atom:updated>2018-04-18T17:12:38.926Z</atom:updated>
            <content:encoded><![CDATA[<h4><strong>Interview with Samuel Woolley, D</strong>irector of the Digital Intelligence Lab, T<strong>he Institute for the Future (IFTF)</strong></h4><p>“My team will get back to you on that, Senator,” and “AI tools.” These seemed to me to be two of the most-used phrases during Mark Zuckerberg’s congressional testimony.</p><p>What struck me the most was Zuckerberg’s tendency to portray “AI tools” as if they are magic bullets for solving the issue of misinformation. But is there a downside to making disembodied “algorithms” the center of our debates around online propaganda? What is lost when we talk only about the <em>things</em>, and not the <em>people</em>? Who designs algorithms, develops them, and optimizes them? What assumptions guide their work? What standards do the makers of algorithms have in mind?</p><p>Back in February, with a hope of finding answers to these questions, I attended a workshop at the Institute for the Future (IFTF) in Palo Alto called “<a href="https://www.eventbrite.com/e/computational-propaganda-a-conversation-with-tech-industry-employees-tickets-42722357722#">Computational Propaganda: A Conversation With Tech Industry Employees</a>.” As a part of his work at IFTF, <a href="http://www.iftf.org/samwoolley/">Sam Woolley</a>, the director of the Digital Intelligence (DigIntel) Lab, gathered technology workers to discuss the role that <em>tech workers</em> themselves play in the global fight against political propaganda.</p><p>Long before all these congressional hearings and hasty regulations, Sam has spent the last five years of his life (from the University of Washington to the Oxford Internet Institute) researching the role of bots in disseminating political false information around the world. The recent result of his work, in collaboration with Philip Howard and Ryan Calo, was published as a <a href="https://www.tandfonline.com/doi/abs/10.1080/19331681.2018.1448735?journalCode=witp20">paper</a> called “Algorithms, bots, and political communication in the US 2016 election: The challenge of automated political communication for election law and administration.”</p><p>Below is an edited transcript of my conversation with Sam about his experiences working with tech workers to find solutions for “computational propaganda.”</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/400/1*PAiEqbxyqC0Bdr4O07KZcw.jpeg" /><figcaption>Samuel Woolley (Credit: Sam’s Twitter Account)</figcaption></figure><p><strong>Roya Pakzad: Sam, I really liked your workshop. It was very unique in terms of involving tech workers themselves in the issue of computational propaganda. So, I’m curious to know, how did you start this initiative and what’s your next step?</strong></p><p><strong>Samuel Woolley:</strong> So, IFTF has a Future for Good fellowship program. Myself along with Michelle Miller — who is a founder of coworker.org [<a href="https://home.coworker.org/">link</a> ] — are both fellows together. Coworker.org helps workers to organize in the digital world because lots of unions have been disbanded.</p><p>The goal is to help workers to be thoughtful, to challenge the status quo and to fight for their own rights and their own voices. Michelle and I thought the problem of computational propaganda or digital disinformation is something that seriously affects tech workers. They also have a real hand in helping to combat it; most of them are not CEOs.</p><p>But they know the platform and they have an interest in making it better, so let’s involve them. The workshop we held here was just a trial run for bringing together tech company workers to lend their voices to this problem and I think it’s going to be part of a larger process of hosting these workshops around the country (and potentially around the globe) and help to rise tech workers up.</p><p><strong>Roya: While many people have been looking for some policy and legal solutions to fight against “computational propaganda,” you took a different path and decided to work directly with technologists. Why?</strong></p><p><strong>Sam: </strong>I have been an ethnographer, so I was most interested in hearing the <em>human</em> perspective on why the technology gets built, how it gets built, by whom and for whom.</p><p>I really wanted to understand beyond the policy parameters of what the technology companies can do to combat this stuff. And it was actually [informed] by understanding the profile of the people who were building these bots and technologies. What I learned is that a lot of them weren’t building bots for political manipulation but they build them to make money. It was kind of illuminating to know that all of these people are struggling to make a living and they were building bots to then rent it out to clients in other countries.</p><p>In other contexts, I did find technology workers who were very vested politically. So I began to understand that there is a dichotomy. On the one hand, you had people who were struggling to make money, and on the other hand there were people who were doing that to make politics. There is a disparity between these two groups.</p><p>I wanted to point out a couple of things: first to underscore the fact that people in developing countries were doing what they had to do to make a living and earn money — to some extent it was similar in the Western wealthier countries. The other thing I wanted to underscore is the ethics of design. Especially when I started to talk to people working at the bigger technology companies (like Twitter, Facebook and Google), what I began to understand was there is not a lot of reflexivity, there is not lots of thought on the part of the engineers when they are building tools on how the tools might be used for manipulation.</p><p>The idea [is] that the people who build technology have a role in production and how the technology gets used (and how it might be misused). The tools — whether social media platforms or the algorithms that dictate what trends get shown — are not agnostic and not free from bias.</p><p><strong>Roya:</strong> <strong>It’s great that you mentioned the concept of ethics for design. Technology companies have been thinking about offering ethical training for their technical employees. What do you think about these types of trainings?</strong></p><p><strong>Sam: </strong>Well, this is my opinion and my guess: I think a lot of the training in those companies was more to do with interpersonal communications in a more routine and old school way. What wasn’t really being conceived of was the way that the tools themselves might be normalized for misuse, like the ways that they would be co-opted by powerful people to try to manipulate the masses.</p><p>I just think it was a lack of foresight in the development of these tools. The companies were thinking about the ethics of gender norms, or the diversity of their hirings, but they weren’t thinking about the ethics that they were baking into the tools themselves.</p><p>When I was a researcher at Google Jigsaw, we would bring in people who all look different from each other to try out the tools and talk about how this tool might be used or misused. I would say research first, then design and launch, rather than design and build, “move fast and break things.”</p><p>For instance, now, I’m working with Jane McGonagall [<a href="https://janemcgonigal.com/">link</a>,] a female game designer, on an ethics challenge toolkit. It basically has three challenges that the people who build technology should go through. There are series of [ethical] questions that they should ask themselves when they build a tool. The idea between me and Jane is to challenge people to think ten years in advance about possible scenarios. The goal is to help people to achieve <em>clarity</em> rather than a concise and exact conclusion [as a result of their new design.]</p><p><strong>Roya: During the past few months, we have witnessed movements by technologists who criticize their companies from within. People who call themselves “tech-reformists” or I might call them whistleblowers. Sometimes their ethical concerns are ignored or even they decide to leave their companies. What was your experience working with them? How can your work empower them?</strong></p><p><strong>Sam: </strong>Currently, I’m collaborating with people like Renee DiResta and Tristan Harris who are thinking about these questions from the perspective of people who work at Google or Facebook. those people aren’t really whistleblowers but are more like tech-reformists. They do believe the technology has to be changed so that it doesn’t manipulate the psyche as much, like Tristan’s idea of “Time Well Spent” [<a href="http://www.tristanharris.com/tag/time-well-spent/">link</a>.] In the past I’ve absolutely spoken with people internal to the companies but I don’t know if I’d call them whistleblowers so much as I’d call them concerned employees who love their work. They were not trying to leak information out but they want to talk about the process.</p><p>I think groups like coworker.org are on the frontline in discussing how to prevent these kinds of miscarriage of justice [tech workers ethical concerns being ignored or getting fired, etc.] Their hope is to make sure workers are heard and have more of a collective voice. We do need more groups like coworker.org but what we do at the DigIntel lab is to give voice to same sort of people but through research, I discuss these things with them and write research to say, “here is what actually comes from tech workers themselves.” My way of giving voice might look different from NGOs like Coworker but we are doing similar work, just through different avenues.</p><p><strong>Roya: And here is my last question: If you wanted to recommend two books for people who are interested in your work, what would they be?</strong></p><p><strong>Sam: </strong>One of them is <em>The New Leadership Literacies</em> by Bob Johansen [<a href="http://www.iftf.org/newleadershipliteracies/">link</a>,] the former Executive Director of IFTF. It’s written almost to business audience but also presents people in government and civil society with notions of how organizations and leaderships are becoming liquid, how they have becoming distributed. It’s really related to the problems we are seeing in today’s world. We are seeing reluctance to face the fact that digital environment means that the world looks more flat rather than like a pyramid in terms of leadership structures.</p><p>For the second book, I think people should read <em>Twitter and Tear Gas</em> by Zeynep Tufekci [<a href="https://www.twitterandteargas.org/">link</a>.] She is a great thinker in this and her voice is really important. As a primer for people who wants to understand what’s going on and what’s happening [in online political movements] I think they should really read her book.</p><p>We wrapped up here. This interview has been lightly edited and condensed.</p><p>This conversation was part of the interview series for my newsletter <a href="https://www.royapakzad.co/newsletter/">Humane AI</a>. I will continue talking with both policy and technical experts in the field of ethics of AI in future installments. Tune in to know their opinions about many issues including AI for social good, History of technology and its connection to AI, Digital Humanities, and much more. To subscribe to the newsletter, click <a href="https://royapakzad.us17.list-manage.com/subscribe?u=9138308bb26620c53a0881c20&amp;id=8152ed9f0c">here</a>.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=706efa1a483e" width="1" height="1" alt=""><hr><p><a href="https://medium.com/humane-ai/designing-for-ethics-706efa1a483e">Designing for Ethics</a> was originally published in <a href="https://medium.com/humane-ai">Humane AI</a> on Medium, where people are continuing the conversation by highlighting and responding to this story.</p>]]></content:encoded>
        </item>
    </channel>
</rss>