RESEARCH

→ View original post on X — @scobleizer

Outstanding AI insights worth sharing regularly

By

@scobleizer

–

10 November 2024 21h14

One of the best posts about AI have seen in months. Please send me more.

Missing Ilya: Talented Researcher and Great Communicator

By

–

10 November 2024 21h09

God I miss Ilya. Not only is he an incredibly talented researcher, he also knows how to explain complicated things simply. And that is an art. I have great hope in ssi https://
x.com/AI_Evolutionis
t/status/1855702769098703359/video/1
…

DeepMath Debate: Transformers Can Solve All Problems

By

–

10 November 2024 20h06

With regard to the debate about the new Math benchmark, Kamil correctly referred once again to DeepMind, who have very clearly stated that they have mathematically proven that transformers can solve all (!) problems. In this respect, it is only a question of time.

AI Development Breakthrough with OpenAI O1 and Path to AGI

By

–

10 November 2024 19h56

The statement that the development of AI is slowing down or coming to an end has now been debunked as BS and proven false.
With OpenAI's breakthrough with O1 and CoT, there is no end in sight to development. Q* aka Strawberry was the breakthrough that brings us to AGI; that's why

OpenAI unlikely to merge GPT-5 and o1 before 2026

By

–

10 November 2024 18h51

You are right, they are going to merge them. However, probably not 2025, as inference-cost for o1 is still too high. That's why I assume there will be one "regular" LLM like GPT-5 (or whatever it's called) and one reasoning model. No merge before 2026

o1’s CoT: Real Thinking vs. Imitation in AI Models

By

–

10 November 2024 18h43

Jasein Wei, @OpenAI researcher has written a great post on how CoT at o1 differs from CoT prompting. And it is clearly different. CoT prompting ultimately just imitates thinking, but doesn't actually do it. CoT at o1, on the other hand, triggers an inner monologue in the model,