Academia’s absurd AI delusion
The collapse of the university platform
I have spent years as an academic, navigating how to produce knowledge in the ways that universities and governments demand. Academia is a weird place. Over time you learn the expensive and inefficient processes of modern university knowledge production (peer review, journal rankings, that no one really cares what you research). It all comes to seem normal.
But today I can only describe my academic experience as absurd - there is a collective delusion that an AI bomb hasn’t just detonated in the middle of this medieval institution. The walls of academia are crumbling, but no one seems to notice.
Academics are calmly walking through the rubble claiming AI either as an opportunity, or a threat to be banned. Some of the most forward-looking ideas are that we can just add AI to the existing platform model of a university with minimal disruption. It’ll all be fine if we make sure everyone is safe, and that the students aren’t cheating. Maybe the university will shrink a bit, but it will all be fine.
If you haven’t been on the internet much lately you may not have realised that:
The cost of producing research is collapsing to near zero. Deep research models now handle the grunt work of academic publication at minimal cost. They produce shockingly good outputs. These models will only get better - they will lie less and they will get better at emulating the very publicly accessible style of academic papers.
The cost of producing and delivering education content is collapsing to near zero. AI can not only produce a syllabus, but write the lecture, write the tutorial notes, produce the in-class activities, turn the content into a podcast (soon a video), and give middle-of-the-night personalised tutoring. I am not making this up.
These are good things - we should accelerate toward them.
But those functions are also the precise things that academics are paid to do. Yes, universities do more than this. A big part of what universities do is give out badges that signal to employers that students have some skills, and they will probably continue to play that role. And maybe the students like human contact.
As we wrote in So, You Run a University?, universities are platforms. Just as Uber connects drivers with riders and Amazon connects sellers with buyers, universities connect students, teachers, researchers, governments, alumni, industry, and others. We counted about 15 groups. Over centuries, they’ve developed large, homogenous institutional infrastructure to facilitate those connections and make those trades happen. And it’s been working.
A radical technological shock that collapses the cost of multiple core functions to zero will trigger a platform restructuring, or collapse. I’m not saying that universities won’t exist, but that if they do then they will look very different. And just because we don’t know what kind of different doesn’t mean that we should delude ourselves that the bomb hasn’t already gone off.
The way I see, knowledge will reside with the ai. Individual learning plans will be created by the ai. Human guide, to keep students on the critical path to success. Marking will be done by the ai. Teachers will be more like Virgil in the Divine comedy guiding the ‘student’ Dante, through the heaven and hell of a subject. It will cost a fraction of what it costs now.
“And maybe the students like human contact.” Previously, yes. The generation raised on screens with less outdoor and peer playtime? Not so much, although perhaps there is an opportunity for academia to help restore the skills lost by Haidt’s “Anxious Generation.” Literally, “the humanities.”