And there's also the question of what data is fed into the system. And how bias is adjusted.
At one point I asked ChatGPT to tell me if there were any issues with a specific credit union, and it ranted in a negative way about (the generic concept) of credit unions. I had to point out multiple times that it can find handful of controversies for the larger established banks, yet it insisted keeping a more pessimistic opinion of unions.
This kind of information is a problem, even when ignoring outright hallucinated false facts (like being casually called a sex offender; recent article about an artist going through this matter), or this example.
At one point I asked ChatGPT to tell me if there were any issues with a specific credit union, and it ranted in a negative way about (the generic concept) of credit unions. I had to point out multiple times that it can find handful of controversies for the larger established banks, yet it insisted keeping a more pessimistic opinion of unions.
This kind of information is a problem, even when ignoring outright hallucinated false facts (like being casually called a sex offender; recent article about an artist going through this matter), or this example.