Last week, I had the honor of delivering a keynote address at the International Conference on Machine Learning in Seoul, titled "What will be left for us to do?" In the talk, I discussed a widespread anxiety: how we should adapt as AI capabilities continue to grow. I was thrilled with the reception of the talk, so I have posted the slides here, along with a lightly edited, annotated transcript. You can also view them directly below on this page, but the online version contains animations, clickable links, and offers a much better overall experience.
I made three arguments. First, AI as Normal Technology is a correct and useful framework for understanding the impact of AI, unless and until some future discontinuity occurs, such as recursive self-improvement. Second, while we should indeed take recursive self-improvement seriously, there is no single milestone in a lab that, once reached by a company, will suddenly put us all out of work. Third and finally, future jobs will look very different and will require a great deal of adaptation. I shared my thoughts on what this future might look like and concluded with a vision of human/AI "co-superintelligence."