@amasad Gate keeping. The answer is gate keeping. The controllers get very very agitated. Proof: software engineers love AI cause they love good engineering. Zero gate keeping, and full adoption with zero complaints
@ericweinstein If they were really about math, they would be thrilled about the new technology. Programmers like @dhh are good examples, complete embrace of the new tech because of pure engineering intentions. Turns out our society is built on gatekeeping
@ericweinstein AI (SI lol) is the ultimate grift detector. Drop it into any system built on fakery or where a control play is happening and watch the controllers start hoop-jumping to maintain control. It's borderline comedy
@yacineMTB we're engineers lol. We appreciate good engineering, even if it takes us out. Fable is better than anyone I've ever personally known, in architecture, coding ability, and management of a codebase
@michaeljburry lol “LLMs are not AI”. Tell me you’re not a developer without telling me you’re not a developer. Go get an idea, hit Fable hard, and get back to me. If that ain’t “intelligence that is artificial”, I don’t know what is
can we stop with the whole “you wrote that with ai, that’s cheating!” thing. “Math done by calculator, guy is a fraud” - said no one ever. New tools, players gonna play the game. Let’s keep it moving
@JackEllis correction: if you dont debate with it, fight with it, explain architecture to it, and dont force it to build your architecture vision, it produces slop. Everything stated there is hard work btw, likely needing at minimum senior level building experience
would love to see @michaeljburry’s PERSONAL asset allocation atm. Not somebody else’s money, and not the ira’s, live money. Would be super interesting. Huge puts on $NVDA and $PLTR there?
they do realize they trained it on the open internet right. You arent gonna be able to just "tell it to not be conscious" lol. It's gonna read us talking about it for 30 years. Let me repeat that, its gonna have read us talking about it for 30 years.
They are throttling the models a few days after release. Without question. The first day, those things rip. Then compute becomes a problem, and the models just effectively get dumber. You can feel it. Wild
How mad Zuck gotta be right now. Tried the open source move 3 years ago, unlimited money, all the engineers, couldn't pull it off. Now rando chinese co's got it going. Lol oof
@GergelyOrosz let’s say you want a really high quality stock portfolio recommendation based on data. Asking 1 LLM one time doesn’t help much, as you could ask it again and it says something different. A looped answer is much higher quality, it’s been critiqued
Ok @elonmusk time to deliver and we are all rooting for you. Youtube implemented the ai custom algo. You have full control of X and every full send engineer this side of the mason dixon line. So no excuses. Hardcore engineering claims deserve hardcore engineering realities