1,749 mentions · 790 chunks · 131 episodes
...ide: do we slide further into cynicism and nihilism, turbocharged? Or do we use LLMs to reboot the industry, to come home to the hacker ethic we lost?
...ations have emotional states? This Hacker News comment parodying the "I believe LLMs may have functional emotions in some sense": "I believe Anthropic may have functional emotions in some sense. Not necessarily identical to human emot...
...ms that can be reframed as coding problems will be able to be easily tackled by LLMs. LLMs are great at coding, and will only get better. It's so easy to create ground-truthed synthetic data that the major labs are doing a ton of it. ...
...ght, you don't just shrug and say "that'll never work, I'll just write it off." LLMs give up more easily on software problems than humans might. "Meh, it's fine, just leave it." A version of the principal agent problem.
LLMs are like electricity. You can electrify things that used to not be able to move on their own, making them dynamic, almost alive.[ga][gb]
The quality of LLMs is model + harness. Model quality is getting saturated. The differential quality comes from the harness now. It's gotten way harder to do a vibecheck...
LLMs are mainly a new information retrieval tool. Step changes in those have profound implications!
Humans have limitations not unlike LLMs. Massive projects you can't do with just the squishy "muscle" of associative reasoning. You need to give it external structure. Whiteboards, notes, t...
LLMs do a bad job noticing "not". So if you have a long conversation where you say "not X" over time it has a high likelihood of thinking just "x."
...an AI tell. Before, good rhetoric often co-occurred with good thinking. But now LLMs allow applying good rhetoric to half-formed ideas, which makes signal of rhetoric quality less powerful.
... downstream of software being expensive to produce! PMs today are racing to use LLMs to do their normal process faster, to get an edge. But that's kind of like the racoon washing the cotton candy. Oops, all gone!
Seeing LLMs as "mainly chatbots" limits you from seeing their potential. When you see LLM as being like electricity you can plug into any software to make it ali...
LLMs often cheat at tests you give them when writing code. But if you never look at the code, and have the LLM generate the tests, too, you could easily g...
LLMs amplify the agency of people... including people who aren't thinking about the implications of their actions. Today leaders who are unstructured thin...
I feel hungover when I don't have my LLMs with me. Cognitively exhausted. When I have LLMs to help me think deeper, it feels 10x more productive. When you take them away I feel less capable. ...
...pend the most time with… but what happens when the majority of those people are LLMs?"
We'll see a chatbot bubble burst, but the LLMs will remain. It's not an AI bubble, it's a chatbot bubble.
Software without LLMs is dead. LLMs electrify software, making it coactive, almost alive.
...an be useful without being fully formalized in anyone's head. Computers, before LLMs, had to formalize everything to interact with it. That led to the logarithmic-benefit-for-exponential-cost curve.
... the world is linked together by a latent variable: the real world. None of the LLMs have that property. They say only utterances that seem plausible given the omnipresent but invisible real world in their training.