Opportunities in AI - 2023
Andrew Ng's latest talk at Stanford University
In my previous two posts, I discussed my thoughts and learnings on AI. Today's post is my notes on Andrew Ng's latest talk at Stanford University titled "Opportunities in AI - 2023". Andrew is a leader in AI and AI education. I have learnt a lot from his AI-related courses, including the recently launched GPT series courses, on DeepLearning.AI, which is founded by Andrew for AI education.
My related posts on AI:
#1 China in the Post-Pandemic Era: A discussion on AIGC and ChatGPT with friends
#2 Revisit AI: My latest learnings and thoughts on AI
My key takeaways from Opportunities in AI - 2023, where I focus more on unconventional perspectives on AI:
AI is not an new industry, but the "electricity" for all industries. It is the foundational infrastructure for various sectors in the future.
AI is not a single tool but a toolbox, which includes supervised learning, generative AI, unsupervised learning, reinforcement learning, and more. The popular generative AI is also based on supervised learning at its core.
The AI toolbox Prompt-based AI is not only for consumer applications but also for developer applications. It greatly improves development efficiency of services, reducing development cycles from months to weeks or even days.
Prompting is revolutionizing AI application development Currently, the most attention-grabbing generative AI may not have the highest commercial value, and this will still be the case three years from now. The commercial value created by supervised learning is the highest and surpasses generative AI. However, generative AI has the fastest growth rate.
Value from AI: Now → 3 Years For AI services, enterprise-level demands have a long tail, unlike consumer services that cater to a single product for all users (I analyzed reasons behind this phenomenon in my previous sharing - Working in the Technology Industry in China). The biggest opportunities for enterprise services are at the application layer, where AI can be used as a plugin in various industries. For entrepreneurs, it is worth paying attention to these “small and beautiful” opportunities. Here are two specific use cases:
AI+Pizza: Using AI to identify if the cheese on a pizza is evenly distributed. This represents a $5 million opportunity. Previously, a solution on this would have required a team of hundreds of engineers.
AI+Agriculture: Using AI to identify plants affected by pests and then spraying pesticides using machines. This improves the safety of agricultural products and reduces pesticide spending.
All potential AI projects (sorted in decreasing order of value) Al creating extinction risk for humanity is wildly overhyped. On one hand, human society has ample experience steering very powerful entities (such as corporations). On the other hand, Al develops gradually, and the "hard take off" scenario (where Al suddenly achieves superintelligence overnight) is not realistic.
Thanks for your reading and I hope you enjoy it. Stay safe and see you next time.
Proofread by ChatGPT.