Banning that would be unjust. Just use a different service provider that sucks less or try to get the company to change the policy.

Are you merely saying that estimates aren't *necessary* (weak claim), or are you saying that estimates are *never helpful* (strong claim)? (I sense a possible motte-and-bailey fallacy.)

Why 19%?

Awesome :) -- I enjoyed listening to the book in December.

Random comment, Ethan: Have you been informed that in your audiobook you call RLHF "RHLF" repeatedly?

Take this 11-question survey before the end of the year!

@vitalikca.bsky.social is a fake account. Vitalik's real acxount is: @vitalik.ca . FYI my followers who follow the fake account: @stefanschubert.bsky.social , @ozziegooen.bsky.social , @charlesd353.bsky.social , @ahappier.world @ericneyman.bsky.social

I don't like the character limit.

My guess is whoever made the graphic put the earliest year mentioned by the person.

E.g Hinton has said 5 to 20 years, so the graphic-maker put 5 years, i.e. 2029. Similarly 2026 is the earliest Amodei has said. Altman has been most ambiguous, but I read him as having a longer timeline than Amodei.

Yes it is, thanks.

This graphic is wrong about peoples' AI timelines.

On Oct 1st, 2024, Demis Hassabis said: "I think there's still two or three big innovations needed [...] to we get to AGI and that's why I'm on more of a 10-year time scale than others [...] I think 10 years is about right." youtu.be/lLgWzsrc1BA?...

Notable Peoples' Views on AI - AI Quotes [Public]

The years in the graphic are not accurate for (at least) Hassabis, Amodei, and Altman. Please don't propagate misinformation.

Sources in spreadsheet. E.g. Hassabis is ~2034 as of 2024-10-01: "...AGI system... and that's why I'm on more of a 10-year time scale... I think 10 years is about right."

Notable Peoples' Views on AI - AI Quotes [Public]

The years in the graphic are not accurate for (at least) Hassabis, Amodei, and Altman.

Sources in this spreadsheet. E.g. Hassabis is ~2034 as of 2024-10-01: "...AGI system... and that's why I'm on more of a 10-year time scale than others... I think 10 years is about right."

Since I've been here for a while… here are some useful features on Bluesky that you might not have discovered yet! 🧵

Hoping it'll be better than alternatives, e.g. due to better recommender algorithms. I already like that 'likes' are public, unlike on X.

People don't bring up neuroscience during explanations of psychological phenomena much in my experience at all, so your 90% experience with a presumably much-larger-denominator-than-mine may be heavily influenced by selection effects.

What Trump means for AI safety A repeal of last year's executive order, for one thing

I 'followed all' due to thinking you did a good job with this post (the only post of yours I've read yet): www.transformernews.ai/p/what-trump...

Writing Doom – Award-Winning Short Film on Superintelligence (2024) YouTube video by Foregone Films

"Writing Doom" is a 27-minute short film that won a $20,000 Grand Prize from the Future of Life Institute. I recommend it. www.youtube.com/watch?v=xfMQ...

Notable Peoples' Views on AI - AI Quotes [Public]

"Notable Peoples' Views on AI - AI Quotes [Public]" docs.google.com/spreadsheets...

I'm collecting quotes on AI timelines, AI risk, AI forecasts, and the impact of AI.

Nice. Social media platforms can add a lot of value. I hope this one continues to grow rapidly and provides more value than its competitors.