Are you folks using Copilot in daily work? I am and yes it’s often hallucinating hints so you can’t really depend on it to provide valid code for you, but sometimes it’s enough to get you unstuck (or help move through writer’s block). But now I experimented to do some old AoC that I never did before and it’s clear to me I’ll have to turn it off for December so it won’t drain all the fun from solving the puzzles. Although I suppose copilot’s hints are the result of AoC 2015 repos being in the training data in GitHub, so hopefully it wouldn’t be so “helpful” on “day zero” puzzles/solutions.

I'm pretty sure the suggestions for older puzzles are only so good because there are many public AoC solutions. When I did 2015 day one recently it filled out the correct solution for part 2 without even any kind of instruction/comments on what the function is supposed to be doing. I don't think it will be by far as powerful for the new puzzles.

I won't be using copilot for AoK - the fun is in the challenge. I'd rather take a day to get to a solution than worry about leaderboards.

I don’t think people still care about leaderboards in the genAI era. The question is more about if using copilot is helpful in terms of having fun and learning things or stands in the way. So far I’m treating Copilot as yet another form of computer assisted coding, like the intellisense/autocomplete/IDEA intentions, and frankly, I never turn off that features in IDE for AoC. Whenever with Copilot I just might.

I see them as quite different tools in that sense. Intellisense/Autocomplete/etc is like an automated golf caddy that selects the right club for you given your location. Copilot points in potentially the right direction and takes most of the swing for you. Or an automated paint/brush selector for an artist vs a sketched outline of your supposedly intended picture. I'd rather do AoC myself than turn it into paint-by-numbers 😉

Fair enough, ymmv. I suppose there can be many approaches to this, all valid for different contexts and different people.