“Prove that you’re human by solving these machine-generated puzzles that machines can solve to give us more data to train our machines, while we run a simple script in the background to verify that you’re human based on your browser’s metadata.”
I had a captcha a few days ago that was something like “click on all the pictures that go with this” with a picture of a saucepan and the grid was food items and other things.
I pointed my phone at it and asked Gemini and almost instantly it said “the potato in the top left and the carrots look the middle of the bottom row”.
While I despise the captchas from a human perspective, the fact that an LLM can solve the challenge isn’t a deal breaker. It doesn’t need to be impossible for a non-human to solve, it just has to be too expensive.
It does certainly shift the equation to stuff like proof of work since a computer can solve it anyway, might as well not annoy the human.
Oh is the person riding the bicycle part of the bicycle this time? What do you want, the corner of the pedal that is just barely in the box, or the out of focus blur in the background that might be a bicycle or it might be a bear. It’s hard to tell.
I have a self bound rule to never do the multi-tile single image ones. Fuck those so much, they’re always wrong. I will refresh for as long as it takes to get a grid of smaller images that may also be hard but I at least have some better grasp over.
I much prefer to see Anubis rather than some bullshit captcha with a grid of AI generated slop that requires 30 clicks to pass.
You mean 30 clicks before it tells you to “please try again” for no reason.
“Prove that you’re human by solving these machine-generated puzzles that machines can solve to give us more data to train our machines, while we run a simple script in the background to verify that you’re human based on your browser’s metadata.”
I had a captcha a few days ago that was something like “click on all the pictures that go with this” with a picture of a saucepan and the grid was food items and other things.
I pointed my phone at it and asked Gemini and almost instantly it said “the potato in the top left and the carrots look the middle of the bottom row”.
While I despise the captchas from a human perspective, the fact that an LLM can solve the challenge isn’t a deal breaker. It doesn’t need to be impossible for a non-human to solve, it just has to be too expensive.
It does certainly shift the equation to stuff like proof of work since a computer can solve it anyway, might as well not annoy the human.
Oh is the person riding the bicycle part of the bicycle this time? What do you want, the corner of the pedal that is just barely in the box, or the out of focus blur in the background that might be a bicycle or it might be a bear. It’s hard to tell.
(Not so) fun fact: The Anubis image is itself AI slop
https://github.com/TecharoHQ/anubis/commit/3e9a93f629e9c5a69bd7fcd6613d27ce9f5231f9
I’ve read somewhere the main dev has comissioned an artist to redo the Anubis mascot.
That one was, but the current one is not
The repo credits a real artist for the anubis image
https://github.com/TecharoHQ/anubis/blob/main/web/static/img/ATTRIBUTIONS.txt
I have a self bound rule to never do the multi-tile single image ones. Fuck those so much, they’re always wrong. I will refresh for as long as it takes to get a grid of smaller images that may also be hard but I at least have some better grasp over.