We started this project while I was on leave at Uber, so a huge thank you to the Uber-Stanford team for pulling this off and making its publication possible Chenyu @chenyuqiu90, Stefan Wager, Inessa Liskovich, Ali Rauh. A special shoutout to @chenyuqiu90 for the hard work, sweat, and persistence that made this a reality.

I remember learning about accelerated gradient in a phd course on optimization in grad school. The professor was very unsettled as to the fact that acceleration was effective, despite having a much less, at least at that time, geometrically elegant theoretical justification... i wonder all the time should good things be pretty, and are pretty things often good?

life rewards action not intelligence

Me: weather in chicago Gemini: Defining research objectives .......⌛️⌛️⌛️⌛️

Chapter 2 of Charting Reality is up. What does a 100 year-old theory have to say about AI inference? Why was modern network engineering theory born in 1909 Copenhagen rather than in the U.S.? Why might a technically sound model fail in practice? I spent a lot of time over the last couple of days reading about the early days of data science and stochastic modeling, and I’m honestly floored by how interesting, and rhythmic, this scientific history was. I could probably have written the whole chapter by swapping out company names from the 1900s with today’s hottest AI startups, and few would even bat an eye! Modeling Congestion - From Telephony to AI Inference github.com/kuangxu/charting_…

Teach not what you know. Teach what you are seeking.

What happened to Cursor? It feels more sluggish and worse than 3 months ago. It's almost unusable.

It's clear by now that AI is going to make routine computer programming obsolete. There's much more consensus that medium term AI is going to vastly outperform the best humans in many intellectual domains and verticals. This is clearly going to shake the foundation of higher education that focuses on transferring "hard" and "standard" knowledge. Some personal speculations for educators and students: - Things are just going to have to change all the time - get used to it. You won't be teaching the same thing for a second time, let alone 3 years. Curriculum, software, materials, everything. The extra burden will be somewhat offset by AI tools that make the updates more expedient. - Fundamentals will stay the same, but how they are applied will need to be updated frequently. Counterintuitively, in the world of cheap AI slops, understanding fundamentals (math, philosophy, language, etc.) can become more of a sought after differentiator. - The focus will shift from mass "hard information" education to smaller, more personal "soft information" education: more laborious, more human to human interactions. At least for the short term, education will increasingly focus on things that can't easily be found in standard texts as sort of a moat until AI becomes more pervasive (or invasive). - The ultimate question higher education has to answer is how and why education adds value to the student's life or career. Bull case: if there'll truly be some sort of universal basic income, higher education may evolve away from about survival (e.g., getting a job) to about enrichment or entertainment (e.g., the joy of understanding and discovery). Bear case: if AI simply creates a pressure cooker of a "race to the bottom," then there will be tremendous pressure for higher education to help students survive (and hopefully thrive) in the new world order; enjoyment will be a second/third order luxury. Unfortunately, I believe the bear case is likely more relevant in the short run. There will be inevitable confusion and pain. Life gives us what we need. Here is to enjoying the journey, for all of us!