▲ | adriand 5 days ago | |
This is a great observation. I think it also accounts for what is so exhausting about AI programming: the need for such careful review. It's not just that you can't entirely trust the agent, it's also that you can't blame the agent if something goes wrong. |