A year after OpenAI’s ChatGPT was launched, we are starting to see the outlines of generative artificial intelligence’s potential impact on our lives.
While the recent Bletchley Park Summit focused on existential risk and misuse by extremists, debate in universities has focused both on technocratic issues (such as automated marking and plagiarism) and more recently on the potential of AI to transform and enhance learning and research – see, for example, the recently published Russell Group principles on the use of generative AI in education.
But there’s a more fundamental question for universities too. How will AI change our economy, and what will this mean for the role played by universities in readying the workforce of the future? The AI Generation: How universities can prepare students for the changing world, a new report for Demos and University of London, draws together what we know about how universities support students’ employability today, and speculates about how this might change as technology advances.
Within our GRASP
The good news is that there is a reasonably strong consensus about the critical employability skills needed today. As computer use and the internet have transformed the knowledge economy, it is not specific technical skills that are most prized. More important are the broader skills of listening to and persuading clients and colleagues, analysing and communicating information to solve problems, and having the ability to manage your own workload, your career and your professional development – the GRASP (general, relational, analytical, social and personal) skills.
Most of these skills are expected to rise in importance in coming years and will remain important in working with AI. Even if generative AI can produce text and images, workers will need critical and ethical judgement to assess what it produces and what it is asked to produce, as well as the relational and social skills to intermediate between technology and humans. So, it appears that the GRASP skillset, with adjustments, will still be relevant.
The less good news is that there is not much evidence on how well universities help students acquire these skills even now, how well these translate into good employability outcomes, or even precisely how they should be defined. Based on a review of academic and policy literature, The AI generation finds that generic “employability” content is unpopular and largely ineffective for students, but material tailored to subject matter and likely career paths much more valued. Active learning approaches – project assignments, collaborative work, peer assessment – seem to work better in developing most employability skills as well as leading to better outcomes, than traditional lecture-based learning.
But it is what happens outside the classroom – work experience, placements, membership of clubs and societies, studying overseas – that has the most impact, whether based on students’ self-assessments, or longitudinal studies looking at graduate employment rates and types. The complication is that these activities are not available to or taken up by all students equally. Students from poorer backgrounds are less likely to participate – partly a matter of cost, but also a matter of feeling, or being made to feel, that you don’t belong. In this way, there is a risk that the very activities that best boost employability are least accessible to those who need them most.
Murky paths
Growing use of AI may intensify some of these risks. At the moment, a degree still acts as a minimum entry requirement for many personally and financially fulfilling careers. Early studies suggest that AI can particularly help less competent workers achieve better outcomes in standardised professional services tests, suggesting that degree holders may need to do even more to stand out from the crowd than they do at the moment, and that lower-level cognitive work may be automated first and fastest.
This also makes the pathway into graduate careers less clear. Currently students who want to pursue careers in most professions have a clear path ahead of them, with employers keen to diversify traineeship schemes. But AI may quickly automate some of the basic trainee tasks – preparing pitch decks and presentations, summarising arguments, working up architectural details, researching legal precedent – calling into question our whole model of professional development.
These are big societal issues, and universities cannot solve them alone. The AI generation recommends research and analysis of what works best in developing employability skills and a more systematic and fair approach to career-boosting activity such as work placements. But universities can also take the lead as civic institutions, convening government, employers, professional bodies and civil society organisations, to consider what AI will mean for our working lives, and what education and training future generations will need in order to thrive.
