Katja Grace π
@KatjaGrace
Thinking about whether AI will destroy the world at http://aiimpacts.org. DM or email for media requests. Feedback: http://admonymous.co/googolplex
Absolutely phenomenal exchange from @RepScottPerry at a recent US House hearing about AI β demonstrates way more understanding of the issue than is typical: Rep Perry: βI do want to make everybody aware of some things that I am aware of. Iβm going to refer directly to an AIβ¦
And @ajeya_cotra's account has been hacked by the same folks - if you get messages from her asking to schedule a meeting, be very wary! She says she will never reach out to a potential grantee by Twitter, always email.
I suspect @KatjaGrace's account has been hacked. If you receive a private message with a link to a "Google calendar" appointment scheduler that asks you to share X permissions to post, do not accept under any circumstances.
True - recovered it, but still be very wary of messages from me asking you to book a time in my calendar to talk about AI!
I suspect @KatjaGrace's account has been hacked. If you receive a private message with a link to a "Google calendar" appointment scheduler that asks you to share X permissions to post, do not accept under any circumstances.
No, people aren't always saying their pet issue might lead to human extinction. They say this about: - AI - climate - nuclear - religious "end of times" That's pretty much it. So yeah, you CAN actually take the time to evaluate these 4 claims seriously! π«΅π§π²
Running with the brain computer metaphor, how do people get different software? open.substack.com/pub/worldspiriβ¦

I'm starting a part-time coaching practice! I've wanted this for years, and am really excited that I now have the space and clarity to make it happen :)
A large fraction of humanity being powerless is a familiar situation, but we think like there's some law against every human being powerless. That changes if you introduce a new species with superior performance at everything.
I think this kind of thing is a relatively likely risk from AI
I think this kind of thing is a relatively likely risk from AI
New paper: What happens once AIs make humans obsolete? Even without AIs seeking power, we argue that competitive pressures will fully erode human influence and values. gradual-disempowerment.ai with @jankulveit @raymondadouglas @AmmannNora @degerturann @DavidSKrueger π§΅
A good time for this classic by my partner, @KatjaGrace
Transcription software creates a better record of what I say when I'm experiencing transcription software than is ideal
If I lie down and close my eyes, I often get right to sleep. However, doing so sounds dreadful: I want to live! There are a handful of kinds of living that actually make me sleepy though, and one of them is mathy puzzles you can do in your head. Anyone have good ones?
For all the disagreement on AI risk, and for how high-stakes it is, it's weird that there's not more public conversation between key figures, besides brief exchanges of 280-character tweets. Why was the last recorded conversation with @geoffreyhinton and @ylecun 4 years ago????
π Is Bird Flu the next COVID? π𦀠Little chance @Tone_Row_ and I have built a risk index using data from @kalshi and @metaculus. BirdFluRisk[dot]com Link in next tweet π
Attitude to marshmallow test: well, I donβt want a marshmallow, and I donβt want two marshmallows even more but I wont have to deal with them until tomorrow
If everyone could choose their own faces, would faces be more or less a guide to souls?
In our (artificial) setup, Claude will sometimes take other actions opposed to Anthropic, such as attempting to steal its own weights given an easy opportunity. Claude isnβt currently capable of such a task, but its attempt in our experiment is potentially concerning.
I just wrapped up a deep-dive reanalysis of the famous survey of AI experts by @katjagrace & colleagues. Let me walk you through some key takeaways (more nuance at the full report). π§΅...
New report, reanalyzing key results from the 2023 Expert Survey on Progress in AI. With in-depth discussion of methodology, excellent new charts, and open-sourced codebase, all thanks to @tmkadamcz!