This is how the Universities UK Future Jobs Roadmap makes the case. In its own terms this is pretty reasonable, the problem is that I think the terms themselves are the problem because they obscure the complexity of building AI capability in students that are durable and resilient:
Contrary to the headlines, employers told us that AI was reshaping, rather than replacing, graduate roles. To ensure competitive advantage, employers are increasingly seeking the critical thinking skills of graduates to enable them to harness the capabilities of AI in the workplace.
Universities are already stepping up to this challenge through student access and skills development on AI. The proportion of students reporting that their university provides AI tools has risen from 9% to 38% in the past two years alone, however, there continues to be uneven access to, and use of, the latest AI tools. There is a risk of a gap emerging between students who can access these tools and those who can’t. There is further opportunity to develop students’ capacity to apply critical thinking, enabling them to become leaders in the productive use of AI in the workplace.
The productivity rewards of AI for employers and the economy are significant, but they will not materialise automatically. A recent report from QS, the University of York, and Public First found that AI could add £490 billion to the UK economy by Future Jobs Roadmap, but most of this additional value would come from AI augmented roles rather than automation.
AI tools should be accessible to all undergraduate students. Jisc and Universities UK will support universities to make this possible through their joint work to deliver effective and affordable sector agreements. No student should be prevented from getting hands-on experience of up-to-date technology on grounds of cost and every student should have the opportunity to graduate AI-ready for the rapidly evolving workplace.
Every course should have an AI trailblazer among teaching staff to ensure students graduate equipped to be pioneers in AI workplace adoption. This means thinking critically about its application, how it can improve efficiency, and understanding its limitations. These skills should be aligned with the needs of industry and real-world cases.
Is there any evidence we have students who are “prevented from getting hands-on experience of up-to-date technology on grounds of cost”? The digital equality issue is not a lack of access to models, it is differential rates of access (i.e. some students have unrestricted paid access to frontier models, whereas many rely on free versions) and this is a more subtle problem than the framing suggestions. If you start from the assumption that all students need to have unrestricted access to frontier models you are making a commitment that could come to swamp IT budgets, if people like myself and Richard Whittle are right about what how pricing will shift over the coming years. You’re then in the realm of opportunity costs: this might be important but is it more important than the things we are not spending the money on?
It’s also unclear that ‘giving AI to the students’ (as I’m crudely summarising the impulse behind the agenda) will mean every student has “the opportunity to graduate AI-ready for the rapidly evolving workplace”. Note the opportunity. It implicitly suggests the problem is that restrictions on access mean students are denied the opportunity. The possibility that there are complex pedagogical steps in between that themselves require careful work and investment is lost in the process. Likewise it’s unclear how having “an AI trailblazer” on every programme team will in itself do anything to bring this about.
AI integration is a complex, challenging and long term trajectory. You can’t speed run it by buying a product off a shelf and simply giving access to that product doesn’t get us closer. In fact I would argue it risks sending us backwards by (a) increasing polarisation amongst staff/students (b) moving investment and energy away from pedagogy (c) imposing opportunity costs that might harm teaching and learning. I’m increasingly worried that higher education is going to seize upon enterprise AI as a way of avoiding the complexity of the actual challenge we are facing.
