This is not actually a technical lesson on AI but it is more of a “Did you know?” type post.

Part of this I have learned in my new job where I am working with AI technology companies and part of it I simply picked up through my own use.

Here goes:

• The most amazing revelation is that AI responses are not always accurate. You have to be willing to check some facts if you ask questions revolving around dates. Such as “When was the movie 1984 released?” Sometimes it will tell you 1984 and other responses may be 1983 or 1985.
• Some responses add fluff content that’s not really pertinent to the question that you have asked of the bot. And, that fluff information may be great to have in some cases but if you are working on a school assignment, that fluff may not be something you want to include. Especially if you are writing to fit a word count. The instructor who made the assignment may recognize the fluff and discount it altogether.
• Sometimes, when asking the AI bot a question regarding words that offer multiple meanings or possibly making a request that needs additional information, you will simply get a response instead of the bot asking for clarification. For example, “What kind of fun things are there to do in Birmingham?” You may get responses for Birmingham, Alabama, when you meant Birmingham, England.
• And sometimes, the AI bot just outright gets the answer wrong through no fault of the prompt. The prompt can sometimes be asked correctly with all the details expected in the question, but the response is still just wrong. You have to be willing to read and review the responses you receive from AI bots.

These are a few of the more popular issues that come up. I have learned quite a bit in the training of AI bots, but I learned quite a bit on my own, like asking about locations and not asking about specific destinations, including a city, state, or even country.

Until next time.