Today, I am going to write a blog about AI and its progress.
This passage is based on the video below, in which Noam Brown, a Research Scientist at OpenAI, presents AI in terms of "Scale".
The incredible progress in AI over the past five years can be summarized in one word: "Scale".
Noam Brown
Yes, there have been algorithmic advances, but the frontier models of today are still based on the same transformer architecture that was introduced in 2017.
He has been engaged in an AI model called "o1", which is a frontier model not only at OpenAI but also worldwide.
He has discovered a new scaling paradigms, complementing the pre-training method used all models including o1, GPT-4 and similar models, allowing AI to improve further and become smarter.
When he was a PhD student, he was involved in developing "Poker AI" and successfully built a model that took time to think before making decisions, defeating poker professionals by a wide margin.
This approach has also been applied to and beneficial for LLMs (Large Language Model), bringing AI researchers a step closer to AGI (Artificial General Intelligence).
That is o1 ー a model that thinks deeply before responding and enhances its planing, mathematical, and coding capability.
Lastly, he stated.
I know that there are some people who still say that AI is going to plateau or hit a wall. And to them I say, want to bet?
Noam Brown