Here’s the rub then: do we count students as individual human beings, or as a human-genAI unit? This question underlies a large portion of the difference between optimism and pessimism around genAI in education. If students are counted as human-genAI units, it seems that knowledge and performance will increase when genAI is present; however, if students are counted as individual humans*, as before, then students may face declines in knowledge and a host of cognitive skills when genAI “does it for them”.
The great thing about doing empirical research is that you can gain insight into phenomena, or in other words, explaining some of the moving parts of a complex system or the whole. In my case, there are concrete actions (e.g., in the form of prompts) participants take to outsource their thinking to generative AI.
From initial inspection of the data, it seems students have outsourced their cognitive functions to genAI across a wide range of tasks. On one hand, participants have indicated their satisfaction with using generative AI, for example, in that it greatly speeds up their work related to their studies. On the other hand, they express worry that using generative AI for certain types of tasks is “making [them] dumber”. This benefit and drawback are inversely related, that is, the faster they get things done using genAI, the less they are engaging cognitively themselves, and the less they understand. For some, this comes to a head in a striking ambivalence: liking genAI yet wishing it were gone.
Despite this ambivalence, students are maintaining and even increasing their use of genAI. In reflecting on just this one thread within the research data, some questions can be addressed, but even more questions arise. This is the beauty of research, of working towards approximating the whole.
Now, I invite you to reflect: How do you engage with genAI and to what extent does this process involve outsourcing cognitive work?
Please stay up to date with our team’s work by following this blog and our research outputs! Happy Summer!
Note: These insights were derived from research conducted across this academic year, with my supervisor Juliene Madureira Ferreira and my helpful research assistants along the way, Mahira Binte Hossain (sessions), Katrina Cirule and Akshit Luthra (third session and interviews). Three rounds of student-generative AI chat log data were collected and then interviews were conducted.
*this is an oversimplification, I do not think we as individual human learners, are ever completely in isolation from our environment, which includes tools, other humans, etc.
Harry Quedenfeld