My own interest in LLMs increased exponentially when, around 18 months ago, I saw a post somewhere that had a guy who wrote his own inference engine in Rust and

My own interest in LLMs increased exponentially when, around 18 months ago, I saw a post somewhere that had a guy who wrote his own inference engine in Rust and demonstrated it running with downloaded open weight models. I tried it out and was quite amazed that even on my laptop (no GPU) I could get an LLM to write Python programs and engage in discussions about Lewis Carroll poetry. It went from "magic thing that needs a data center of unobtanium GPUs to do questionably useful stuff" to "thing that does useful things even on a regular computer". There's plenty of sand on the planet and clever people (and AI) figuring out how to do more work with less sand and power, so any argument that AI is going to cost so much that it won't be usable, seems just preposterous.