Post

LLMs strike me, strictly from a development perspective as a pretty shit tooling for anyone who has written plenty of their own tools across multiple domains. But, lots of devs never bothered or bother to learn how to write tooling.

Let's not confuse tools with modules. Different paradigm, and a tool never exists inside the end product. A precast concrete form isn't a tool, the machine that makes whatever precast form you want with dimensional inputs is. It's an interesting set, but low stakes tell us nothing about LLMs.

High stakes r what a bunch of us play for, and right now its high stakes game is trash. And it's competing against decades of tool history that already exist and re-exist to chunk out entire deployable/functional applications for hot customization to exact business requirements in a button press.

Here the thing: we have a very valuable system for that right now - but it's really hard to learn to use effectively I saw LLMs compared to guitars recently: you don't pick up a guitar, twang it for five minutes, get crap results and announce "guitars don't work! All hype no substance"

0 reposts0 likes

Hard to use effectively hill is where systems die, and where LLMs will die. The same goes for tarot cards. Because there are other forms of AI that do work and have been built in which will produce deterministic results every time. Can someone make beautiful text with an LLMs like a 🎸, maybe.

And if the music industry was to rest on the back of guitar pluckers and not my friends who are producers, mixers, sound engineers, markets, etc, it would be dead by next Tuesday. All of which are using deterministic systems to squash the variability of the guitar pluckers to make it monetizable.

The squashing of variability isn't cheap, and early arguably LLMs time don't make it cheaper. Having done studies of entertainment financial models where you swap in genAI tools for artists, I can assure you it's not cheaper.

Was before these companies started raising prices (see Cursor) because having a sheep dog that also supposedly hunts quail is a dog that does neither well. Better to have two separate dogs. We'll see if LLMs can ever specialize with high accuracy, but without a digital sheep dog policing it ymmv.