🔥The 9th Round of Easy Loan, Earn $40 Reward is in progress❗️ ⏰ Promotion Period: January 15th - Feburary 15th, 2025 👉 Register now and check more details at gate.io/campaigns/358

What happens if AI just keeps getting smarter? Our latest release, made in partnership with ControlAI, explores how we go from today's chatbots to AI vastly smarter than all of humanity. With corporations racing to build God-like AI, is humanity on the verge of extinction?

And it gets worse: Dario has also warned AI could cause literal human extinction, an outcome that would eliminate at least two-thirds of white-collar jobs

"The AI industry has more lobbyists in Washington and Brussels than the fossil fuel industry and the tobacco industry combined." (full conversation below) Max Tegmark and Robert Wright

Writing about extinction risks from AI is hard. On one hand, I believe that this is an emergency, with a strong possibility of hitting a point of no-return in 2-5 years. On the other hand, I believe the best way to tackle it is to plan methodically and then calmly do all that

I am working on a little side-project. Nothing ground-breaking, just a nice way to copy-edit my articles using Claude. And there was a bug (shown on the image below) where the suggestion is off-by-one. It didn't matter much in practice in my specific workflow, but if I wanted

Great podcast with the ever insightful Max Tegmark , making beautifully clear a lot of the most important challenges we face! Give it a watch!

I have many times fought against Markdown. It always misses something that I want and it's a pain to extend. So, I created a small extensible alternative, thanks to a general command syntax. Here is a small video!

Category Theory is one of the biggest nerd-snipes I have encountered. My goto example of abstraction for its own sake, way past its ability to cleanly explain things. When it is used, it makes things less clear. How many people have been turned away from FP because of monads?

Thinking that mechanistic interpretability is the key to understanding AI safety is like thinking that neuroscience is the key to understanding political science.

My advice on (internet) writing, for what it's worth

What may be a just process we should follow to eradicate an entire species? Does it take into account sentience, harm caused to people, our aesthetics? Is there even a just process? Should we set up "It's obvious!" as a precedent for something that large or pause and reflect?

37 UK politicians backed the campaign on AI extinction risks ControlAI briefed No10 1000 UK citizens contacted their MPs 2500 US citizens contacted their senators 2 partnerships with Youtubers and more in the pipeline Soon, 50000 people on the newsletter Proud of ControlAI!

When people say current models are aligned, this is much more out of a lack of specification (the same a slow buggy react app is "ok") than out of the expectation that agents built on this are good for us.

I try quite hard to be honest, and I am very glad to read this. Thank you Dr Lawrence Newport. Well-founded trust (as opposed to tribal favouritism) is rare, and may only be built with time. I believe it is the main way to do great things together over long-time horizons.

I have just created a website where I put my tinkering projects. gabe.love The latest project is GERFECT, a small web-app that helps one develop perfect pitch using melody triggers (melodies that starts with target note, making it easier to identify and learn)

Democracies make it hard for individuals to have too much power. Many take this personally and get frustrated: they want to have a large individual impact. This leads them to what I call The Ideological Spiral. More on cognition.cafe/p/the-ideologi…

I have created a small webapp to train perfect pitch. It uses melodies as mnemonic devices to help identify notes. If anyone wants to try it and give me some feedback, I'd love it! perfect-pitch.app