From behind the scenes: #ICML2026 comms chairs share what they were up to. Blogs, social media, interviews, press.
First and foremost: a fundamental philosophy of transparency and openness. We hope you felt the same.
Check out the post to learn more: blog.icml.cc/2026/08/10/what…
I had the honor of giving a keynote at @icmlconf in Seoul: “Moving from Behavioral Guardrails to Principled Agency.” 🚀
Simple guardrails are no longer enough. We have entered the "teenage years" of AI.
Read the compressed highlights on X:
👇 👇 👇
I had the honor of giving a keynote at the International Conference on Machine Learning in Seoul last week titled “What will be left for us to work on?” I addressed the widespread anxiety about how we should adapt as AI capabilities increase. I was thrilled by the talk’s reception, so I have made my slides available, annotated with a lightly edited transcript: cs.princeton.edu/~arvindn/ta…
I made three arguments. First, the "AI as Normal Technology" framework is a correct and useful as a way to think about AI’s impacts, unless and until there is some future discontinuity such as through recursive self-improvement. Second, even though we should take recursive self-improvement seriously, there is no milestone that companies might achieve in the lab that will suddenly put us all out of work. Third and finally, jobs of the future will be radically different, and a lot of adaptation will be needed. I shared my thinking about what this might look like and ended with a vision of human/AI “co-superintelligence”.
That's a wrap! We hope you had a fun, productive, and idea-filled Gangnam-styled #ICML2026! It's been a pleasure posting with you -- stay tuned for more content. We wish you a safe trip home, and see you next year!
The last part of the welcome remarks is recognizing all the award winners. Check out the blog post for more details on the winners and the process!
blog.icml.cc/2026/07/05/anno…
Many were curious or concerned that choosing a dual policy of LLM use would result in disparate outcomes between the two policies. The punchline: the policy choice did not affect the score distribution or decisions in a statistically significant way.
The dual policy for LLM use featured two policies for LLM use, conservative or permissive. All reviewers were matched with a policy consistent with their choice.