We have now concluded our crowdsourced study scrutinizing political videos on YouTube during the 2022 Brazilian elections. More here: news.exeter.ac.uk/faculty-of-env…

This is the AI risk that scares me the most...

How do algorithms shape your news feed on FB & IG? We dug into @meta’s transparency updates and... we can't tell. Researchers Claire Pershan The Law Technologist Jesse McCrosky found Meta's updates do not give a full picture of how news feeds are shaped. ⤵️ mzl.la/3qfnveM

A very cool study that makes a great complement to the RegretsReporter User Control Study we conducted with Mozilla foundation.mozilla.org/en/youtube/use…

No Perfect Solution to Platform Profiling Under Digital Services Act The #DSA comes into effect for large online platforms today. How to address personalization-related obligations is a challenge, say Jesse Jesse McCrosky and Claire Pershan (Claire Pershan): techpolicy.press/no-perfect-sol…

Generative AI is increasingly integrated into our day-to-day lives. At Thoughtworks, #responsibletech principles are at our core. I share thoughts and best practices on deploying AI sustainably in a new article for InfoQ infoq.com/articles/carbo…

Want to learn how your organisation can build the policies and structures needed for real #responsibleai? I'm excited to share a new Thoughtworks whitepaper on AI governance that I co-authored with Ryan Dawson - Software for Humans and Meissane Chami. Check it out now! thoughtworks.com/insights/white…

PS: I´m jealous I didn´t come up with this slide

This is a fascinating example of AI bias mitigation in practice. AI models will never be unbiased, so mitigation is appropriate, rather than quixotic efforts to "debias" them. But is this the best way to handle mitigation? What else could we do? I'm curious what you think :)

The agenda for Decarbonize Software 2023 is ready 💚 Hear the latest announcements in #greensoftware and how practitioners are improving culture, knowledge and the application of green software to achieve software sustainability! ➡️ grnsft.org/decarb #decarb2023

I've had many opportunities to speak on responsible AI recently, and I always end up talking about Goodhart's law. As such, I'd feel remiss not to share Adam Mastroianni's excellent analysis of the concept: open.substack.com/pub/experiment…

Join us at #Decarb2023 on Nov 16! Tammy McClellan & Jesse McCrosky explore greener AI solutions. Get your free pass 🎟️➡️ grnsft.org/decarb #greenAI #greensoftware #decarb2023

Important insight.

Vitamin D offers perhaps the most poignant example of the problems of observational studies in all fields. Almost every retroactive obs study shows strong link between Vit D levels & health outcomes. Almost every RCT of Vit D supplentation shows little to no benefits. via Alex Telford

the most important factor for a genuine ethics/responsible AI work within an AI company/big corp is the willingness of the company/corp to create a safe space for critical work (including self-critique) and dissent

With the @Mozilla Foundation, I'm beginning a program of research studying the impact of generative AI on elections. Our first experiment has found that OpenAI is not enforcing their policies around generation of political advertising. foundation.mozilla.org/en/blog/chatgp…

With Mozilla, I'm kicking off a program of research evaluating tools for detecting AI-generated or -manipulated content. Here's why this is important. foundation.mozilla.org/en/blog/naviga…