LLMs haven't really gotten "smarter" - but the tools we use with them have

[00:00] LLMs haven't gotten a lot smarter, but [00:02] they kind of haven't needed to because [00:03] like the the Chrome around these tools [00:05] has gotten increasingly more [00:07] sophisticated as we introduce hooks and [00:09] then and then skills. And now it's like [00:11] um what I'm working on now is is tools [00:14] to increase the verifiability of of that [00:17] what I'm asking for is actually what is [00:20] produced. And so like now we're in this [00:23] like window, right, where it's like [00:25] we're talking about a universe of less [00:26] than 12 months. I think a lot of people [00:29] are walking around with a lot of [00:30] confidence that they know exactly what [00:32] this means. Market's doomed. They're [00:34] just like, you know, they're they're [00:34] drawing a line that goes up into the [00:36] right or down and to the right and [00:37] they're saying either like we're never [00:38] going to have programmers again or all [00:40] this stuff is going to explode because [00:41] like the the fundamental economics of AI [00:44] companies don't make sense. So, it's all [00:45] going to explode and we're going to go [00:46] back to the old good old days any day [00:48] now, right? Like people are walking [00:50] around like they're really sure what the [00:51] future holds. I think we all need to [00:53] have a little bit more humility and we [00:54] need to take a fresh look at what these [00:56] tools are actually capable