Linus Torvalds was more stubborn than his AI assistant
The assistant did much of the tedious work on an Intel graphics memory bug: adding debugging code and analyzing the results. Several times, it also declared the problem impossible and suggested writing a report instead.
Torvalds kept pushing. The investigation continued, a fix emerged, and he let the AI write the commit message.
His explanation, collected by Simon Willison:
I suspect those things have been trained by people who may not be quite as stubborn as I am.
A capable helper, with someone else still deciding when the investigation was finished.