Very interesting, how do you validate the correctness of the generated visuals? And that a previous lesson, say 5 videos ago ensures that the concept being explained now had its prerequisite covered?
I assume you could do a topological sorting of the content to ensure prerequisites are covered, but I'm curious on how automated this could be.
Haha I love the concept!
What if you allowed smaller local llms loaded in the browser as an unpaid intern tier? You wouldn’t charge an hourly wage for using them but you could charge for the “skills” you teach them so make them useful. Just a suggestion, don’t take it as prescriptive.
I think whenever a new technology comes out, there will always be some new jobs to:
- build (LLM research engineers)
- integrate into existing systems (LLMOPS and AI engineers)
But I think besides those obvious ones, there will be lots of new combinations of LLMs and other technology that we won’t be able to predict, but in hindsight seem like it was destined to happen.
Genuinely curious on the types of projects you use Haskell for! I’ve been thinking of learning it beyond the lightweight treatment I got during my CS degree.