In my post China in the Post-Pandemic Era in March, I shared my discussions with friends on AI. There have been many updates in the AI field recently, such as Sam Altman testified before members of a Senate subcommittee. I have also tested more interesting products, like AutoGPT. In this post, I would like to share some of my latest learnings and thoughts on AI. Any corrections or further insights on this topic are welcomed.
Competition and Opportunities
1 - Will AI lead to a future where OpenAI dominates or a more prosperous ecosystem emerges?
I believe it will be a case of "one superpower, many strong players", divided between open-source and closed-source approaches. OpenAI is poised to become a super-giant, but open-source will foster a more thriving application ecosystem that can meet diverse needs. Stablediffusion vs. Midjourney is a case in point. The reason open-source can find its place is that it doesn't need to achieve perfection, and certain scenarios are better suited for open-source or are likely to be fulfilled by open-source solutions in the first place.
2 - Will China miss the AI wave?
China may be slower, but it will not miss out. This is because of the open-source nature of AI. Correspondingly, NVIDIA has also customized GPUs for China. If it cannot be blocked, why not make money on it?
3 - Before hardware undergoes generational shifts, there will be significant opportunities for smartphone manufacturers. Imagine the intelligence of your Siri progressing from an equivalent level to a human at one year old, to your current level, and surpassing you in intelligence.
4 - There will be numerous opportunities in productizing open-source capabilities, such as vertical services based on AutoGPT.
Long-Term Impact on Humankind
1 - AI will train humans in reverse. For instance, Go players evaluate their own moves based on the actions of AlphaGo. When 99% of problems can be solved directly through AI assistants like Siri, will human patience and problem-solving abilities progress or regress in addressing the remaining 1% of challenges?
2 - While AI brings convenience to work and life, I believe it is important to maintain a clear and cautious boundary between AI's integration into work and life. For example, determining which use cases warrant AI and which workflow stages should incorporate AI. Your willingness to outsource certain abilities to AI depends on which abilities you, as a human, take pride in and consider precious. When in doubt, I believe the right course of action is usually the more difficult one.
3 - Which areas are less susceptible to AI disruption? 1) Domains focused on human entertainment or personal enjoyment, where there is no unified evaluation system and no continuous progress, such as art. However, this does not diminish the importance of these fields. 2) Disciplines with limited available data.
4 - In the post-AI era, individuals will have increasingly more idle time to fill, and there will be a growing need to fulfill idle attention.
5 - Chat interfaces and prompts are just one type of product feature designed to facilitate interaction between AI and humans. They carry on the legacy of the browser era, but they will not be the sole means of AI-human interaction. In the future, interaction methods and the definition of interaction itself will differ greatly from what we are familiar with today.
Thanks for your reading and I hope you enjoy it. Stay safe and see you next time.
Proofread by ChatGPT.