Great article, in particular your look into the accessibility aspects of CAPTCHAs. I recently had my own encounters with Microsoft's new CAPTCHA that's nigh-impossible to clear, I included a screenshot and an explanation of what's obviously wrong with it under "A New Captcha" at the end of this article in case you haven't experienced it before. If you haven't - challenge yourself! I'd love to hear your take on this one. (Googling "microsoft account captcha" brings up result after result of humans who simply cannot get through it!)

I've also, repeatedly, experienced a specific and culturally jarring visual CAPTCHA, which inspired this meme: South Africans have always called traffic lights "robots", so showing a South African an image of a traffic light with the text "I am not a robot" is actually quite unsettling!

Considering that LLM AIs like ChatGPT have APIs and are now more capable of solving simple language and logic puzzles than most humans, I don't feel like those are the way to go. Having said that, between my personal experiences and your article, I've come up with a suggestion for a more usable and reliable alternative to the current CAPTCHA mechanism: a single service that one can sign up for that's dedicated to determining one's non-robot status. We'd only have to establish ourselves as human once, using multiple well-designed challenges, and then connect those to our various email accounts (just like Gravatar).

And then we'd never have to do these absurd challenges again.