That said, in the case that the author brought up here I agree that web searches shouldn’t be performed. I personally don’t like having my main agent search the web (mostly for preserving my context) so I’ve restricted the web search tool to a dedicated web search subagent.

- Student A started with Scratch, but within 2-3 lessons I convinced his mother to switch him over to Python. He wasn’t my best student, but he was very open to trying things and had a fun time coding. Plus it was really great to see him go from zero to competent within a short period of time. Personally I think it’s 50/50 that he continues coding on his own, but I hope the skills I taught him (not the exact syntax) will stick with in some form or another

- Student J is far more interesting to me as he’s not interested in making games or coding. Rather, he prefers making animations and really likes the Scratch interface. While his animations aren’t the highest quality stuff I’ve come across, I really admire his attitude of making one animation per lesson and trying all kinds of stuff vs sticking to just one thing.

Personally I think scratch is a decent start for teaching kids to code, but I wouldn’t recommend it for that purpose. If you want to teach your kids to code, I think it’s best to start with a “real” language. That said, I think it’s a very good introduction for animation and graphical projects without all the complexity of a game engine.

If researchers at a well funded institution like MIT are seriously considering replacing hiring undergrads with LLMs, I can just imagine how researchers at schools with less funding are open to it. I never did research as part of my undergrad (something I regret, but I likely wasn’t in the headspace for it back then) but my friends who did view it as a core part of their education and deeply helpful for future opportunities. I really hope schools come up with a policy to help discourage this.

I’m really curious about trying this myself. Recently I’ve been handing off tasks for agents to complete on their own more and more. While their work is acceptable I’ve found them to not only take a long time to complete it, but they often add too much complexity and tests.

When I got my M3 Pro MBP, I got it with the intention of eventually installing Linux on it. In hindsight I seriously underestimated the amount of effort needed for even partial Linux compatibility, which makes me all the more grateful for the work Asahi contributors have put in! I'll still be keeping macOS installed for iOS dev work(which I plan on doing soonish), but other than that all I really need is a web browser and Ghostty with good battery life.