1,598 mentions · 615 chunks · 117 episodes
English specs can now be "compiled" to runnable code by LLMs. When we make programs, we compile the source code down to an executable binary. But then we make sure to keep the source code, so when we tweak it i...
Having a background in things like sociology gives a leg up for understanding LLMs. Most tech is understood best with a straightforward math/computer science frame. But LLMs are a cultural technology. They are best understood throug...
LLMs didn't train on images, but pictures. An image could be any random bit of noise expressed as a 2D array of pixels. A picture, in contrast, was a thin...
...stem pulls to an extreme, rigid, brittle, centroid. A heat death of the system. LLMs just imitate, following the innovation that humans that came before have done. In a world where we use LLMs for more things, how can we make sure tha...
...he top models. We can now take an IQ of 90 for granted with cheap off the shelf LLMs… and it will only get cheaper. What happens when you take IQ-90 LLMs for granted, and assume it's too cheap to meter?
One of the reasons that LLMs appear to be so resiliently good at frontend UX in modern patterns is because that code isn't challenging in a programming sense, it's a lot of boile...
... five year old, they just confidently distill an answer on the spot. Not unlike LLMs! We're all inherently creative. Confidently answering a question on the spot, distilling an answer that is plausible within the context of everything...
...t outcomes from stupid components. What happens when we allow smart things like LLMs to use them?
LLMs write shitty software quickly. They struggle to write high quality software, even with lots of scaffolding. To work in Serious development, the softw...
It's amazing how useful large context windows are in LLMs. It's barely been a year since we had to deal with miniscule context windows of 4k tokens or so. It was like living in the stone age, can you even im...
...or. In that world, we'd have LLM-powered aggregator chatbots, but no way to use LLMs in other applications. A recent report said that something like 75% of OpenAI's revenue comes from ChatGPT. All it would have taken in this alternate...
LLMs are excellent teachers. They can patiently engage with our questions, helping you learn the material. But imagine trying to learn German and using an...
...r system to make that UI applied to a different set of your data. A party trick LLMs are quite good at.
Some people critique LLMs as being just like BitCoin: a massive energy hog. However, there's a key difference. In BitCoin, the energy use is the point. The price of computatio...
The assumption that Chatbots are the killer app for LLMs presupposes a centralized, necessarily one-size-fits-none system. When you centralize, you have to have a one-size-fits-all policy or approach, and g...
...software that fits everyone like a glove. This is now possible because software LLMs mean software in the small is no longer expensive to write.
LLMs only do superficial pattern recognition, but they can do it incredibly robustly. They are amazing at superficial absorption of patterns. But if you b...
...a tamagotchi. Anything with a face that you can talk to. Pond scum has no face. LLMs and other emergent algorithms are closer to pond scum intelligence than human intelligence. But LLMs put a face on pond scum intelligence. You don't ...
Lots of people are talking about how LLMs might change how large organizations work. I think LLMs will almost certainly have a big effect on how organizations work. But I don't think it will ...
One-ply thinking: LLMs will make navigating bureaucracies and paperwork easier. Multi-ply thinking: LLMs will make it so that bureaucracies processes get even more labyrint...