I have been using the paid for version of ChatGPT lately. Here are some initial thoughts.
Research. I have used it to verify my understanding of historical events, and to optionally add to that understanding. It has been very useful in this respect. However, it is not infallible. The problem seems to relate more to missing elements rather than erroneous information. If I prod it .. “what about x” .. then it will include the missing item(s).
Writing Style. This is undoubtedly more polished than my own. However, apart from very occasionally using a short paragraph, I do not use longer pieces because: (a) they do not correspond to my writing style; and (b) the message that they are seeking to portray generally differs from my own.
Code. I have used it to write a small amount of code in VBA, relatively successfully. It was mainly a case of stepwise refinement, i.e. you need to check that the results are correct rather than just assuming that they are. I note that Claude seems to be the go-to Chatbot in this area.
Chatbot use of my material. My website is not a commercial undertaking, and so I have no great objection to the use of the information so long as it is attributed. Question – how do they know if it is reasonable to trust it?
Very Simplistic Comparison of Chatbots
I asked them to give me an overview of my website:
- ChatGPT gave the most detailed response (possibly because it knew me?)
- Gemini arguably came second, just ahead of Claude
- Co-Pilot and Perplexity both produced shorter summaries.
I am sure that they will all improve over time.
It would be useful to hear the views of others.
BK .. This and That 