We’re announcing the Inverse Scaling Prize: a $100k grand prize + $150k in additional prizes for finding an important task where larger language models do *worse*. Link to contest details: github.com/inverse-scalin… 🧵

And remember: This is just the public results!

We are looking to hire Research Scientists for our Long-term Strategy and Governance team at DeepMind! Deadline Aug 1. Please forward and apply: boards.greenhouse.io/deepmind/jobs/…

$653,392 for this line of research which is incredibly high risk and which was not at all useful in response to the coronavirus pandemic we just suffered, to a person whose conduct has raised serious integrity, transparency, and conflict-of-interest questions.

Is there an easy way to get all papers on Open Review like an RSS feed?

The AGI tried to align me first.

I basically agree, in my days a SJW the thing that the people we were dealing with really missed about us is we were incredibly scared of them. Current EA have a responsibility to the future of the movement to stop seeming like a powerful rich cult with crazy views.

This is a gross simplification. Here is a more accurate phrasing: "There are ~300 AI researchers primarily motivated by AI x-safety, and ~40,000 who are not." It is not accurate to say those who are not are working on "making AI more powerful". This reads like a political ad.

Adversarial Policies Beat Professional-Level Go AIs abs: arxiv.org/abs/2211.00241 project page: goattack.alignmentfund.org

It's notable to me how many longtermism critiques focus on philosophy, and how few focus on what longtermists are actually doing, which is good stuff like preparing for pandemics. IMO it's some update towards talking about x-risk rather than longtermism (on strategic grounds).

We have free GPUs to support your research! Thanks to a generous donation from Jeff Dean, we have 28 GPUs to give away. Here is the full inventory: (Nvidia GeForce GTX) Titan: 10 1080: 2 Titan X: 6 Titan Z: 1 Tesla: 9 You are welcome to apply for 1-4 for research use.

AKA GPT-4 is going to be here and doesn't need advanced prompting techniques

I honestly have no clue if FTX did bad things, but if they did do bad things, I would like to say now that I think it’s bad that they did bad things. Also, I think it’s good that they did the good things they did.

If the FTX situation has left your grant-funded project in the lurch, happy to see if Mercatus Center can help, DM me.

I’ve had “withdraw btc from blockFi” on my to do list for about 2 weeks now. Sensed it was coming but got lazy. Don’t procrastinate! Secure your bag.

... at least I don't see it much outside of EA. I see EAs worry about it sometimes! I know I do. If EA in the end is net negative, I strongly bet it was not because of FTX but rather because of shortening time to AGI

Guess I can finally put on my cv I trained a 52b GPT arxiv.org/pdf/2211.09110…

You should retweet this paper. arxiv.org/abs/2211.16480

I'm looking for a freelance designer to help out on some Our World in Data projects. You'd be working on data visualization and art for upcoming articles. Flexible hours, competitive rate. Is this you or someone you know? Reach out via DM

Google was about Borg not PageRank, and OpenAI might be about Triton not GPT-3. Companies have a tendency to turn algorithm moats to infrastructure moats, which are much more durable.