Re: FreeDOS caveats

On Fri, 26 Apr 2024, Ivan Shmakov wrote:

On 2024-04-25, Ben Collver wrote:

>

It looks as though some effort was put into composing this
message.  Nice work, assuming it is human generated.

>
When working time and again as a part-time university
lecturer, I'd typically feed bits of my students' works to
Duckduckgo to detect plagiarism.  Makes me wonder if this
approach can be adapted for detecting AI-generated texts.

Granted, it is not university level work or students, but what I do is to just compare students work. The AI-people more often than not, hand in the exact same source, same spelling, same variable names. That is enough proof for me to not pass them.
But sophisticated cheaters rewrite, and then it is of course a bit more difficult. In that case I do manage to catch them during code reviews when they cannot explain what they do or the algorithm.