CAPTCHAs Becoming Useless as AI Gets Smarter, Scientists Warn

Ok the main reason that bots are a problem is they use a lot of resources without giving any value to the peoples providing the resources, most modern bot detecting systems use heuristics to determine if you are human or bot, which generally work reasonably well.

One thing I've found where it does not work very well are situations where I am very good and productive at what I am doing because then I evidently am less like a human and more like a bot.

Probably the heuristic should indicate an expertise level on the platform that could say something like User X has been on this platform 200+ hours probably they are a little bit more efficient than our baseline user profile.

That said this scheme if it only told you this machine is registered to exactly one human user then that combined with the heuristic could be enough to not make bots such a problem. Because it really isn't a problem that a bot happens to go to 3 pages a minute and scroll through the whole page and copy a bit of text, the problem is a bot that hits hundreds of pages. A bot that is as efficient as an average to slightly above average human is not a problem.

The heuristic should be something like

definitely a bot (did something no human can ever do, typed at a speed of 10010 WPM typing speed. Bot bug in keyboard emulation) maybe a bot (percentile chance) with checks that do not just increase your chance of being a bot, but also decrease it - this user wrote their username in the user panel, with one common misspelling (rearrange letters) at an acceptable typing speed, then before they went to password they backspaced to the misspelling and spelled it again correctly - decrease bot chance by 2.