There’s one valid use-case for LLMs: when you have writer’s block, it can help to have something resembling an end product instead of a blank page. Sadly, this doesn’t really work for programming, because incorrect code is simply worse than no code at all. Every line of code is a potential bug and every line of incorrect code is a guaranteed bug.

I use an LLM with great success to write bad fanfiction though.

You can save time at the cost of headaches, or you can save headaches at the cost of time. You cannot save both time and headaches, you can at most defer the time and the headaches until the next time you have to touch the code, but the time doubles and the headaches triple.

What anon called “awkward silence” was the entire class being persuaded by his strong opening line.

Neither LLMs nor ICMs are AI in any sense of the word, is my point. LLMs happen to give the illusion of intelligence because of their language-based nature, but they’re not fundamentally different from ICMs.

Image classification model isn’t really “AI” the way it’s marketed right now. If Google used an image classification model to give you holiday recommendations or answer general questions, everyone would immediately recognize they use it wrong. But use a token prediction model for purposes totally unrelated to predicting the next token and people are like “ChatGPT is my friend who tells me what to put on pizza and there’s nothing strange about that”.

Any other newspaper should know better. The Washington Times doesn’t want to know better, because “I didn’t know” is a very convenient excuse when you publish Nazi shit every day.

Google Lens already did that though, all you need is decent OCR and an image classification model (which is a precursor to the current “AI” hype, but actually useful).

Time to read Neuromancer again. It’s such a perfect cyberpunk novel, it’s weird that it didn’t get a movie adaptation with terrible CGI in the 90s.