In the past, whenever I saw an AI button in a program, I wanted to at least try how it worked. And almost always, it performed quite poorly and was incapable of solving my real tasks.
Over time, I accumulated so much of this experience that now, when I see yet another AI feature in some service, my main interest is simply to close it as quickly as possible and use the core part of the service where AI hasn’t interfered yet.
The key is not to forget to occasionally check whether AI is still incapable or has actually learned something because negative assumptions don’t lead to progress.
Top comments (0)