Very cool, if true

Also, wouldn't it be sensible to do some research into how accurately AI marks homework *before* you tell teachers to use it to do that? Here's some research from February which shows that its agreement with examiners was only 0.61 (where 1.00 would be perfect agreement).

Large language models are only accurate by accident. They're like the old game show "family feud," which gave points for guessing the most popular answers to survey questions.

The difference is, the show wasn't asking factual questions like "what's the capital of Russia?" or "find the hypotenuse of this triangle," they were subjective or specific to the individual. Things like "where do you eat breakfast?" or "where did you spend your last vacation?" An LLM might do well in that game.