Here's an example that we did in class:

Our university provides access to both Gemini and Copilot. So I asked
both to write a summary of the FreeDOS Project, including its history,
key dates, and some contributors. I figure I know that history pretty
well.

I don't remember which was which, but one claimed that I have a
computer science degree from the University of Minnesota.* The other
said FreeDOS 1.0 came out in 2006 and 2.0 came out in 2025.**

__

* Wrong: I have a physics degree (minor in math) from the University
of Wisconsin River Falls. I earned my MS (technical writing) from the
University of Minnesota.

** Yes, FreeDOS 1.0 came out in 2006, but we haven't released "2.0"
yet. FreeDOS 1.4 was in 2025. I think AI got that because I did a lot
of interviews leading up to "1.4" and I described the next version as
"1.4 or 2.0 or whatever we call it" because we hadn't decided on the
label by the time I did those interviews.


On Sat, Oct 3, 2026 at 1:50 PM Jim Hall <[email protected]> wrote:
[..]
> I have used this assignment for the last 2 semesters (i.e. 3rd time
> with it) and AI is getting *better* but it's still not great. I find
> it interesting when the students realize "the closer I look at the
> essay [because they are evaluating it] the more it looks kind of
> shallow and generic." And some students find that AI has claimed some
> wrong "facts" (usually correct in the "big picture" but wrong in the
> small details) and probably one of the 3 citations is either plain
> wrong (made up) or has an error (usually the wrong date or author).
>
> So I'm not ready for AI to do my work for me. :-)


_______________________________________________
Freedos-devel mailing list
[email protected]
https://lists.sourceforge.net/lists/listinfo/freedos-devel

Reply via email to