LLMs Will Replace 8-Track Duplication Engineers

But now we have LLMs. They’re also a black box, and if you throw enough tokens and context at them, they’ll out-perform humans. They won’t be able to tell you how they did it, either.

This isn't an interesting finding when we consider that a simple bruteforce search is almost certainly computationally lighter than feeding the data to a language model and guarantees optimal results.

The interesting thing is that LLMs can do this at all, I guess.

This is utterly fantastic and makes me grateful to those anonymous engineers because I remember getting really frustrated with the blank space on a handful of 8 tracks in my mom’s Skylark back in the day.