Every instructor should know by now that AI detectors don’t work and have high false-positive rates against non-native English speakers 👇 & so a lot of people are turning towards less formal methods of identifying AI cheating. Most of those introduce bias as well. To take two common things I have seen people do:
1) Punishing people who write obvious AI essays means you are merely punishing bad AI users not AI use overall, since good AI use has more natural sounding results and is not easily detectable
2) Punishing people whose writing improved the most, because they are likely using AI, means that people who were previously cheating (using essay writing services, etc) don’t get punished, and you just focus on people who gained the ability to cheat. What do you do about previous cheaters?
As instructors, we need to figure out how to work with students to reconstruct the value of homework in an age of AI, rather than just punishing them, especially given that cheating was already common.
