Who's Who of AI

Jonathan Stray

66 trust researcher @jonathanstray.bsky.social · 880 followers
Knowing things is a solved problem. Getting along is not. Working on AI, media, and inter-group conflict @CHAI_Berkeley. Got here from computational journalism.
What they're sharing

Articles & links

Hi! Ready your paper, very interesting. Thought about it, and I'm not sure I find the computational complexity proof convincing because it bites only for learning all possible distributions D -- human-like D is plausibly easier. I found this, makes argument more rigorously. ar…

Barriers to Complexity-Theoretic Proofs that "AGI" Using Machine Learning is Impossible arxiv.org
AI Weekly's analysis →
  • Guerzhoy argues van Rooij et al.'s 2024 proof that AGI via machine learning is intractable rests on an unjustified assumption about data distributions.
  • The same proof structure, applied consistently, would show ImageNet classification is intractable, yet that task demonstrably works.
  • Three barriers block any such proof: defining human-like behavior precisely, accounting for inductive bias, and specifying relevant data subsets.
Read full analysis →
View on Bluesky · ♥ 0 ↻ 0 ↩ 1 · 2 from the directory shared this · 13d ago

Well, there are examples in the paper and the dataset is open source! github.com/humanCompati...

github.com
View on Bluesky · ♥ 0 ↻ 0 ↩ 1 · 12d ago
Their own posts

Recent commentary

Humanity's ability to know, reason, judge, and act well is the foundation of science, democracy, crisis response, & management of AI itself. AI poses serious risks to that foundation. New paper on epistemic risks by 30 experts calls for attention and proposes solutions. Link in thread.

View on Bluesky · ♥ 34 ↻ 19 ↩ 3 · 29d ago

What could it mean for an AI to be "politically neutral”? And can we measure it? New paper + dataset. We propose a definition that applies to any type of conflict on any topic: a neutral response should maximize approval on both sides of an issue, while keeping that approval balanced. 1/🧵

View on Bluesky · ♥ 4 ↻ 1 ↩ 2 · 32d ago

Seeing a flurry of evals and startups promising to test the mental health effects of AI. Literally all of them test what the model says in various conditions... none of them measure actual outcomes on actual people. A big gap, fixable with privacy-preserving experiments.

View on Bluesky · ♥ 5 ↻ 0 ↩ 1 · 43d ago

Is this a good logo for the GreenEarth feed? It's a healthier, user-controllable, open-source, LLM-powered, transparent feed we're building -- now in alpha testing. Try it? Tell us what you think! bsky.app/profile/did:...

View on Bluesky · ♥ 3 ↻ 1 ↩ 0 · 2d ago

I want to make sure AI doesn't incite human conflict. It's sometimes hard to explain what I do, but that's the core of it -- it won't happen automatically. And we're making progress! Both theoretically, and in field experiments that test how AI alters human relationships.

View on Bluesky · ♥ 2 ↻ 1 ↩ 0 · 7d ago

AI safety typically assumes one well-meaning user. I'm working on the case where two of them are at war.

View on Bluesky · ♥ 2 ↻ 0 ↩ 0 · 24d ago

Hello I am in Montreal for the week! Anyone interesting in AI safety I should meet here?

View on Bluesky · ♥ 1 ↻ 0 ↩ 0 · 8d ago
Their network

In Jonathan Stray's orbit

Center = Jonathan Stray. Left = members they follow (green edges). Right = members who follow them (blue edges). Top = mutual follows (orange edges, slightly larger). Drag any node to reposition; click to open that profile.