[rsynnott]

> To clarify my analogy, GenAI is definitely a "problem" rather than a particular solution, and as such I expect it to have longevity.

Hrm, I'm not sure that's true. "An 'AI' that can answer questions" is a problem, but IMO it's not at all clear that LLMs, with their inherent tendency to make shit up, are an appropriate solution to that problem.

Like, there have been previous non-LLM chatbots (there was a small bubble based on them a while back, in which, for a few months, everyone was claiming to be adding chat to their things; it kind of came to a shuddering halt with Microsoft Tay). It seems slightly peculiar to assume that LLMs are the ultimate answer to the problem, especially as they are not actually very good at it (in some ways, they're worse than the old-gen).

[falcor84]

Let's not focus on "LLM" then, I agree that it's just a step towards future solutions.