if you’re walking round SF does it feel like the early days of covid, where it’s clear what’s on everyone’s mind?
Can somebody explain to me what concrete thing happened in the last 48 hours that explains the fact that I’ve seen 57,246 vague posts like this one
Why are some people in the tech scene in a daze the last few weeks? I'll tell you. I've been spending an ungodly amount of time with the new AI tools that are only a few weeks old, testing out the creation of teaching materials for my kids and my students.
This stuff is better than anything I could ever come up with. Scarily so. And these are all very recent leaps in capability (<2 months). The tools have gotten so much better at working on jobs for hours and days and weeks. They are tenacious, smart and get complicated jobs done flawlessly.
What does education even mean anymore... when you have models that can recall everything humanity has ever learned with near perfect precision? Many of our education systems are elaborate schemes to screen people by ability and work ethic for scarce jobs or posts. But what does any of that mean now? It all seems pointless, and the jobs might not even exist soon. I have no idea what my kids are going to do.
Are we going to let our minds atrophy and leave the heavy lifting to the machines? We still have the edge on creativity and new knowledge creation, but how long will that last?
Just on a lark, I was able to create, in about 2 weeks, and a hundred million tokens, an entire medical school curriculum. 450 lectures and notes, 16000 figures. Multiple rounds of error checking show that the work is 99% flawless. Going to release and open source it soon. Normally this takes years of work from hundreds of humans. Instead an agent swarm from Claude Code was able to do it in days.
It is both exhilarating and demoralizing to see this happen. No knowledge work is safe. The economy will feel this soon. But unlimited knowledge is freely available to everyone.
This is something for which there is no precedent in our history.
Some examples of the output I've been able to get:
What llm did you use to create the medical graphics?
Nano Banana 3 Pro
First Anthropic and now OpenAI have just released new products that were built mainly by AI over the course of a couple weeks, and we may be on the verge of an explosion of new technologies as LLMs begin to recursively self improve at a rate much faster than what was possible when human researchers were the bottleneck.
Opus 4.6 and codex 5.3 and their ability to summon swarms of carefully controlled task-oriented agents. Except for a principal engineer to occasionally remove the agents from doom loops and just serve as a cyber-janitor, entire teams can now be put to pasture.
No comments:
Post a Comment