Someone asked me the other day how they can best go about identifying AI-generated media. "I don’t know" was my honest answer. To be completely truthful, I’m not even sure I can. With that, I am confident in my ability to distinguish essays written by students from AI (and have done so in the past). I can also pretty easily spot dodgy-looking fingers and feet in many AI-generated images. But, of course, such examples are not the case for all AI-generated media. Moreover, what about when AI advances to the point where it can overcome these little "nuances"? What happens when it’s capable of "fooling us" a majority of the time? Maybe I’ve already been duped and I don’t even know.

Perhaps a better way of looking at this question of identification is through a pre-emptive lens. That is, let’s assume that we will be duped on a regular basis (if it’s not already the case). I don’t think being able to "spot a fake" is necessarily the right approach. Consistent with a couple of pieces I wrote a while back for this blog, regarding spotting fake news and why we fall for it, it’s more about the decision-making process regarding accepting what to believe and the critical thinking necessary for such decision-making than it is about identification per se. Indeed, AI has been a topic of discussion within critical thinking research recently (e.g., Dumitru & Halpern, 2023; Eigenauer, 2023; Saiz & Rivas, 2023)—for example, with respect to its potential effects on human attention and decision-making, along........

© Psychology Today