At the recent TED AI conference, Noahm Brown, a research scientist at OpenAI, sparked enthusiastic responses from the audience. In his presentation, he introduced OpenAI's newly released o1 model, emphasizing how this model transforms decision-making across industries through "System Two thinking."
Image source: The image was generated by AI, provided by the image licensing service Midjourney
Brown has been involved in the development of the renowned poker AI "Libratus" and the diplomatic game AI "CICERO." Now, he hopes to see artificial intelligence not just as a tool, but as the core of innovation and decision-making.
Brown mentioned in his speech that over the past five years, the development of AI can be summarized in one word: scale. Although there have been many technical advancements, today's cutting-edge models still rely on the transformer architecture introduced in 2017, with the main difference being the scale of data and computational power.
He emphasized that it's time for a paradigm shift. AI needs to move beyond mere data processing and into the more deliberate "System Two thinking." This concept originates from psychologist Daniel Kahneman's book "Thinking, Fast and Slow," referring to the slow, thoughtful way humans approach complex problems.
Brown also shared an interesting story: during the research on Libratus, he found that having the AI think for 20 seconds in a poker hand resulted in performance improvements equivalent to expanding the model size by 100,000 times! This discovery was shocking to him and he believes future AI development should focus on this System Two thinking.
With the release of OpenAI's o1 series models, AI will be able to process information more meticulously, particularly suitable for complex tasks such as scientific research, programming, and strategic decision-making. Brown pointed out that the o1 model achieved an accuracy rate of 83% on the International Mathematical Olympiad tests, significantly outperforming the previous 13% by GPT-4o. This performance makes the o1 model particularly important in data-driven industries.
Additionally, Brown emphasized the broad application prospects of the o1 model in the business sector, such as improving decision-making processes in healthcare, energy, and finance through System Two thinking. He even jokingly asked the audience how much they would be willing to pay for a new cancer treatment, further highlighting the value of slow thinking.
Despite the high operating costs of the o1 model, Brown still believes that for businesses requiring high-precision outputs, the investment is worthwhile. He concluded his speech by stating that AI development is at a critical juncture, and future success will depend on how effectively System Two thinking is scaled.
Key Points:
- 🧠 **System Two thinking is crucial for future AI development, enhancing decision quality.**
- ⏳ **A 20-second thinking time yields better results than a 100,000-fold increase in data.**
- 💡 **OpenAI's newly released o1 model demonstrates outstanding performance in multiple fields.**