Weakest of the Major LLM Chatbots in My Experience
Community Review for Microsoft Copilot
My Take
In my use of Copilot and other Chatbot LLMs, Copilot seems consistently the least well trained, and the least capable of giving me a detailed and direct answer. I often find it gives me partially incorrect information and that I have to adjust my prompt several times. I wouldn't recommend it for finding teaching resources, accurate information, or helping you to plan in the classroom.
If you do use it with the paid version, you probably have better features and immigration if you work in a Microsoft environment. It seems very dependent upon whether or not you have a paid license and how much you're integrating into the Microsoft suite generally. But I would always check objective information that Copilot gave me.
How I Use It
There are a lot of LLM chatbots to choose from these days. I tend to run the same prompts through several at the same time to see what I get and how they compare and contrast. Some excel more at creativity type of tasks some are better at the cold hard facts and analysis.
The best example I have of this kind of comparison for Microsoft Copilot is when I asked it to help me with Outlook for Mac. I had a question about a feature that seemed to be indicated in Microsoft Help sections, but they were all geared towards Windows users. I asked copilot if a feature existed on Outlook for Mac and then had to alter the prompt at least a dozen times and still got a wrong answer.
I put the same prompt in Google Gemini and got the correct answer in the first shot. It seemed as though Copilot wasn't even correctly trained on Microsoft help articles.
This is not my only example of this kind of dilemma with Microsoft copilot, it is just the most salient one.