Lack of Collaborative Learning in ChatGPT Limits Professional Use:A Case for Shared Intelligence

In asking ChatGPT to do things as simple as write a BAT for windows, I have encountered egregious errors. In one case ChatGPT thought it made a mistake on line two, identifying the Desktop forgetting what it programmed in Line 1. Asking it about whether it “Learns” from its mistakes it replies that within our conversations it does but NOT across conversations. This is a VERY serious flaw. If ChatGPT is EVER expected to do serious professional work, either helping programmers, or helping doctors with diagnosis, or engineers with calculations needed, we can NOT expect each indivual to have to TRAIN ChatGPT themselves for their purpose. That is completely unsustainable. Moreover the issues of privacy, verified accuracy, repeatability, and security CAN be addressed adequately. I can help with this, if asked. Until then ChatGPT is, as I have told them NO MORE THAN A TOY and no one should be willing to PAY to use it.

This topic was automatically closed after 23 hours. New replies are no longer allowed.