This article offers a fascinating perspective on the development of human intelligence, like looking at it from a godlike viewpoint. It's worth reading carefully.
My notes after reading it:
1. While AI can quickly overcome certain problems requiring advanced intelligence, such as studying ancient texts, new materials, and protein folding, advancements in fundamental physics seem less straightforward, perhaps due to limitations in observational methods and groundbreaking theories?
2. At least currently, consumer-grade AI lacks sufficient expertise. Therefore, you have a way to assess your competitiveness: if your knowledge and experience in your field are not stronger than AI's, or if you cannot assess the accuracy of AI's answers to your professional questions, or if you never find any errors in your interactions with AI, then your professional skills are no longer competitive. Others can replace you with an agent system at ever-decreasing costs.
3. I feel that "intelligence per unit price" should be a useful metric in the short term. I wonder if there's a suitable benchmark for measuring this. In other words, if today you can pay 100 yuan to have an AI with an IQ of 120 solve a problem, in the future it might only cost 10 yuan. The cost of acquiring intelligence continues to decrease, eventually reaching a bottleneck (similar to chip manufacturing processes hitting physical limits). This is also very useful for observing AI development from a macro perspective.
4. Could interstellar migration be the major race in the second half? This way, we won't be confined to considering issues within this blue dot, because "how to survive" will become a crucial challenge.
