Yes but you need to setup quite a bit of tooling to provide feedback loops.
It's one thing to get an llm to do something unattended for long durations, it's a other to give it the means of verification.
For example I'm busy upgrading a 500k LoC rails 1 codebase to rails 8 and built several DSLs that give it proper authorised sessions in a headless browser with basic html parsing tooling so it can "see" what affect it's fixes have. Then you somehow need to also give it a reliable way to keep track of the past and it's own learnings, which sound simple but I have yet to see any tool or model solve it on this scale...will give sonnet 4.5 a try this weekend, but yeah none of the models I tried are able to produce meaningful results over long periods on this upgrade task without good tooling and strong feedback loops
Btw I have upgraded the app and taking it to alpha testing now so it is possible
I have a similar game interest. Something based on internet sourced data, e.g. gpt-3 based npc dialogs, trained off current news or something else that's never ending... something like kenshi with the complex mechanics of dwarf fortress, combined with NPCs that can very realistically mimic conversations
It would be an open world, but not pvp...maybe you support co-op but I think the novelty would lie in single player combo'd with a very dynamic cause and affect world
I had this same thought reading it. Anecdotal but my experience with prolonged sleep deprevation was a noticeable decline in my design capabilities.
I'm actually refactoring some of the code/sub systems I created during that time now and some solutions are very wacky...creative, but a super weird way to approach it.
Sleep deprivation might not have been the only contributor, but there a noticeable change in our code base during that time and I distinctly remember worrying about my sleep