In some challenges, the GPT-4-based model triumphed. In others, it failed. How do you know when to count on it?

To mark the launch of our BBC micro:bit - the next gen campaign, we've put together a coding quiz to test your knowledge. See how you score on the eight questions below that are based around the ...

ChatGPT achieved 22% accuracy, while the top human coder in the study achieved 47%. "We saw a couple of human coders perform better in almost all cases than the tool," said study lead author and ...