2 comments

[ 0.22 ms ] story [ 12.7 ms ] thread
> AI adoption raises homework scores by 18% and reduces completion time by 30%, but lowers monthly exam scores by 20% within six months. High-stakes entrance-exam scores fall by 18 and 24%, with the full penalty emerging only after about two years.

In Stephenson's "Diamond Age", an AI tutor-in-a-book helps a girl learn. Some [1] have suggested that current AI already can function like this book, which the study above disagrees. So is the problem that current AI needs to improve as a tutor, or is the model of using an AI as a tutor flawed?

[1] https://www.real-morality.com/post/diamond-age-ai-tutor

This particular system is predicated on people actually doing the homework.

There were threats on 'cheating', but the system is failing at detecting the new 'cheating' mechanism.

I use quotes because what is being cheated at is the certification, not education. Mixing them into the same pipeline was always a mistake.

BTW: A tutor is a person. An AI is a search engine.