Loading…
29
Hot take: I blew off that '63% of teens use AI for homework' poll, then a cashier in Tulsa set me straight
I saw that stat going around a few weeks ago and flat out dismissed it, because the survey only had 400 respondents and the questions were vague. Then I was buying a phone charger in Tulsa and the cashier, who is 19, mentioned she uses ChatGPT to check her essays before turning them in, and both her roommates do the same. That one chat made me go back and read the actual methodology, and it turns out the sample was weighted by region and age, so it is not as junk as I assumed. Now I want a bigger sample, but I have stopped calling the number fake. Does anyone here know how they screened out the kids who just guessed?
1 comments
Log in to join the discussion
Log In1 Comment
colescott2h ago
A buddy of mine teaches 11th grade English in a decent-sized district and he got curious about the same thing, so he asked his five classes to fill out the survey a second time from memory with no internet allowed. The kids who had guessed the first time mostly switched their answers, and a few straight up admitted they had no idea what the questions meant. He said the ones who really used AI could name the tool and describe what they asked it, which is a pretty easy tell. From what he saw, the screens probably rely on follow-up questions and consistency checks, since a faker cannot hold a story together across five items. If the poll did not include a "never used it" option that still let you finish the survey, the guessers would pad the number.
4