Much as I dislike it, ChatGPT has changed the way we do things and look for things. Initially I thought that it was a fad/phase and when people would realize that it gives incorrect information mixed with correct info they would stop using it but that doesn’t seem to be the case. A couple of days ago we were having a discussion on worms and how much protein they have in them in a group chat of friends and Surabhi was tried to gross Anil out, instead of getting grossed out Anil asked for a recipe he could use to cook the worms. Immediately Surabhi went on ChatGPT and asked it for a recipe but it refused to give it stating that it is against their policies and might be disturbing to see. Before ChatGPT she would have searched on Google for the recipe and gotten it (I did that in a few mins after I saw her comment). The a few days later another friend commented similarly where they couldn’t find something on ChatGPT so decided to give up instead of searching via a search engine.
Other people have stated that they use it for tone policing of emails to ensure they are professional. Personally I would recommend The Judge for that as I had stated in my review of their site earlier this year.
The problem I have with ChatGPT is highlighted by the following quote shared by @finestructure (Sven A. Schmidt) “Whether it did it correctly I don’t have the expertise to evaluate but it was very impressive sounding.”. The way GPT works it gives information in a very well crafted manner (and that is super impressive) but the fact that it can have errors or it hallucinates from time to time makes it useless for detail oriented work for me. If I have to verify the output generated by ChatGPT using a browser then I might as well use the browser directly and skip a step.
I have screenshots of so many examples of how ChatGPT/Bing/Bard hallucinate and give wrong information. I think I should do a follow up post with those screenshots. (I have them saved in a folder titled AI nonsense 🙂 ).
– Suramya