What happens to learning when students decide for themselves when to use AI? TODAY: AHA Seminar with Osbert Bastani (Associate Professor, UPenn), from chess academies to AI tutors, on what his field studies reveal about AI in real classrooms. Link: zoom.us/j/91647096870

You check the label before you eat it. Why not before you talk to it? The ImpactBench AI Nutrition Label scores how AI models affect people, from sycophancy and bias to creativity and social connection. Try it! impactbench.media.mit.edu/nu… #ImpactBench #MITMediaLab

Food comes with a nutrition label. AI should too. ImpactBench AI Nutrition Labels turn 800+ behavioral metrics into an at-a-glance view of how a model affects people, traceable to the underlying transcripts. Explore: impactbench.media.mit.edu

Most AI benchmarks measure what models can do. ImpactBench measures what AI does to us. 800+ behavioral metrics. 48,540 conversations. Ten models. Undesirable behaviors avoided in only 53.1% of evaluations. Explore: impactbench.media.mit.edu

AI is tested on how smart it is. Almost no one tests what it does to the people using it: our thinking, our wellbeing, our relationships. On October 7, that changes. Follow along. #AI #MITMediaLab #HumanFlourishing

A new tool for measuring helpful & harmful behaviors of AI. Tomorrow. October 7th, 2026 #AI #MITMediaLab #AHA

AI is the new monolith: built by us, still alien. Fluent answers invite trust but reveal little about the machinery, or what relying on it does to our judgment. At @medialab, we have developed a nutrition label for Alien Intelligence. More coming soon. #AI #MITMediaLab #AHA

Not so cute and fluffy, after all? 🧸 Our MIT research shows users struggle to see what's beneath the interface. Do you know what's in your AI? We've built a way to look inside and measure when AI supports human flourishing, and when it puts it at risk. Stay tuned.

Book Launch! In "Artificial Intimacy", MIT's Sherry Turkle asks what happens when a synthetic pretend empathy feels like enough. Hosted by Advancing Humans with AI. Launch talk + Q&A 🗓️ Thu, Oct 8, 5–6 PM 📍 MIT Media Lab, 3rd Floor Atrium media.mit.edu/events/aha-sem…

The gap between what people say they value and what they actually share online is where algorithmic amplification lives. BONUS SEMINAR THIS FRIDAY: Ziv Epstein (postdoc, MIT Schwarzman), hosted by Prof. Karrie Karahalios. Link: zoom.us/j/91647096870

What do you teach a student about AI when no checklist covers what they'll actually run into? THIS THURSDAY: AHA Seminar with Daniella DiPaola (postdoc, MIT Media Lab) on virtue ethics as an approach to AI literacy. Link: zoom.us/j/91647096874

Mainstream moderation systems routinely misread Black cultural expression as toxic. What happens when a community writes its own rules instead? THIS THURSDAY: AHA Seminar series featuring Rishi Balakrishnan (founding engineer, Blacksky Algorithms) Link: zoom.us/j/91647096870

What's your real relationship with @claudeai ? Our AHA team collaborated with @AnthropicAI on Reflect, Claude's new beta feature that shows your AI habits and helps you stay aligned with your personal goals. Check your Claude settings (desktop) to explore your own patterns!

Whether AI helps or harms us is a matter of design. At #IUI2026, AHA Director @PattieMaes called for AI that serves human flourishing: interfaces that enhance our potential and strengthen our agency. Because there's no point in advancing AI if humanity doesn't advance too.

We optimize AI decision support for accuracy and speed. But what if those aren't the right goals? THIS THURSDAY: Krzysztof Gajos (Harvard) on expanding the success criteria for AI-powered decision support 🔗 Join us: zoom.us/j/91647096870 #AHA #MITMediaLab #AI #HCI