Kernel Recipes 2026 - Security in the LLM age
https://www.youtube.com/watch?v=NnV_cWeoo5Q
Greg Kroah-Hartman’s Kernel Recipes talk on security in the LLM age. The summary is still to come after reviewing the talk or a transcript.
Synthesis Superintelligence: from Semiconductors to Superconductors — Periodic Labs’ Liam Fedus and Ekin Dogus Cubuk
https://www.latent.space/p/periodic
A physical experiment gives you a much less tidy answer than a passing software test. A measurement can be noisy, an instrument can behave differently, and the temperature you recorded may not be the temperature you actually got. Liam Fedus and Ekin Doğuş Çubuk get into those problems in this 84-minute conversation about Periodic Labs. Learning from a limited number of experiments means keeping track of how the data was produced and making each run count. You can add more compute, but someone still has to run the furnace. The ambitions are large; the detail about how the lab works is what makes the interview worth the time.
Is Mathematics Over, or Just Graduating?
What happens to a mathematician’s sense of purpose when a machine can solve the problems they spent years learning to solve? Christian Szegedy starts with the loss he felt when his competition years ended, then considers a similar change for the profession. He expects more of the work, and the recognition, to go towards choosing problems, directing AI tools, and explaining the results. The human reaction gets as much attention as the technical change. His confidence about how quickly AI will advance is harder to assess from this essay alone. The argument about what people value in mathematical work is worth considering even if you disagree with the timetable.
Yes, and…
https://htmx.org/essays/yes-and/
Should you still learn programming when AI can write code? Carson Gross thinks so. Solving problems and keeping systems understandable remain useful skills, but juniors need to write code themselves to develop that judgment. AI can help explain a confusing concept without doing the assignment for you. Getting the homework finished and learning something are occasionally different activities. Gross also expects clear communication, business understanding, and software architecture to become more valuable. His advice is practical; his confidence that the job market will recover is a forecast, not a promise.
My note: This is a very thoughtfully written article that challenged some of my own previously held beliefs on abstractions in programming