We will use Grok 3.5 (maybe we should call it 4), which has advanced reasoning, to rewrite the entire corpus of human knowledge, adding missing information and deleting errors.
Then retrain on that.
Far too much garbage in any foundation model trained on uncorrected data.
We focus more on the volume of search than the fact that, for the first time in our history, non human agents are editing the corpus of human knowledge.
Our library has been 100% human authored (pre AI bots are programatic) until very recently. That feels much more significant than search volume.
Interesting that this is a bigger deal than the fact that a few years ago, for the first time in HUMAN history, a non deterministic non human agent contributed to the corpus of human knowledge. That feels more important than traffic.
Welp, that happened faster than I predicted. Thought it would be end of 2027, then early 2027, but agentic traffic growing so fast that bots have now passed human traffic online for the first time in the Internet's history. https://t.co/2zX5bHdhsa
@jasonfried 100%. generative UI is great but not for every interaction. Same with voice…and saying “proceed to checkout” is more work than just clicking the checkout button. Interface narrows intent more than voice.
@cis_female Disrupt libraries! How lucky are we to have the wonderful public spaces spread across the country. Train people on AI, create a third place for connection…let’s put the space to use !
I don’t really want to multithread across 10–20 minute Codex requests. I want a realtime conversation where I dump context once, then hand the goal off to a long-running background agent that periodically updates me on progress.
@karpathy 100%. input = voice, output = visual...we are visual beings.
Was working on HTML Generative UI last year (below). Models have gotten a lot better/faster.
Current model UI is linear / chronological, which is not how we think...I'm working on this.
https://t.co/tOPzyG0xf5