I’d be curious if there is a good way to combine the LLM with human feedback. E.g. if you’re on Facebook or whatever, the LLM could send that information to someone to have them “hold you accountable with a bit of their soul”
kaiwilliams
Karma: 2
Ideally this is someone you are in regular contact with (a partner/roommate) that can give you more time, but maybe it’s embarassing because you told them you want to spend less time on it.
Another approach (which is the one I take) is to have the person who made the screen time passcode write it down on a sheet of paper that I keep in my wallet. Then, if I need the code, I can ask anyone else to put it in. This only works if you’re able to keep the rule of not looking at the piece of paper, though
I can’t tell you the exact source, but I saw an OpenAI person tweet that this isn’t the actual CoT