The web contains a plethora of unauthenticated, unfiltered information and most students lack the critical skills to penetrate this mass of undifferentiated material. In short, traditional notions of quality in higher education seem to be abandoned in the move to web 2.0 learning.
Rennie, F. & Mason, R. (2008). E-Learning and Social Networking Handbook (free access via Internet Archive)
I feel like we’ve been through this argument before, and have largely accepted that:
- some content on social media is poor
- some content on social media is excellent
It seems reasonable to think that we can apply the same reasoning to generative AI:
- some content it generates will be poor
- some content it generates will be excellent
However, I’d suggest that there’s another variable in play here, which changes everything.
Frontier AI labs are intent on building artificial general intelligence. Let’s set aside whether or not this is possible; for now, it’s clear that they are striving for this.
It’s therefore in their interest for AI-generated content to be excellent (by whatever definition of ‘excellent’ you want to apply).
And, there is no equivalent condition being imposed on human creators of content on social media. In fact, the incentives are terribly misaligned with social media and content creation (because, advertising), creating a forcing function that drives attention-hacking.
This shift raises important questions for educators and institutions. How should we teach students to critically evaluate both AI- and human-generated content? What new skills are required in a world where the motivations behind content creation are so divergent?
There are also risks: if the incentives for AI development shift—toward commercialisation, bias, or other priorities—the quality of AI-generated content will suffer in the same way that we’ve seen social media spiral into dark places.
It may be that our role as educators is to help navigate not just information itself, but the forces shaping its creation and dissemination.