Universities should be cautious about “hyper-granular profiling” of students using artificial intelligence, according to UCL’s vice-chancellor, who fears there is a risk of stifling their ability to grow and change.
Michael Spence said that while new technologies may be useful in teaching technical skills, they could not replicate some of the human processes university learning relies upon.
Many have claimed that the advent of large language models and chatbots can result in more tailored learning for students by better analysing their strengths and weaknesses and knowledge gaps.
But the UCL president and provost told the Reinventing Higher Education conference, hosted by IE University and Imperial College London, that this risks reducing students to data points.
“When I spent 20 years of my career at Oxford…you got a first or a second-class degree or a third-class degree, and that was more or less what we could say about a young person,” Spence said.
“Then we moved to these ridiculous grades where everybody pretended that 67 was something different to 66 and 65 and it really meant something about a young person.”
“The great thing about the old bands was that it gave people plenty of space to develop over time…Personalised education will take that to the extreme.
“Some company or somebody or other will know exactly what your weaknesses and strengths are…and therefore it’s inevitably depersonalising in that young people grow and develop at all sorts of different rates.”
Spence warned that “hyper-granular profiling can become depersonalising, fixing students to metrics and shrinking the space to grow and change”.
He added that while interacting with AI systems “feels personal”, people were in fact disclosing information “to a whole system of people who build the machine, control the machine, know what comes out of the machine, determine what goes into the machine”.
“So I’m actually in a relationship where I am self-disclosing to a vast crowd, a corporate mass of some kind, whether that be a company, an educational system, a government or whatever it might be,” he said.
Campus resource: AI over-personalisation can hinder learning
“That’s quite different to the context where I hire a tutor for my children.”
Spence said that even where AI performs well in subject-based instruction, it cannot replace the “inherently interpersonal and value-laden” nature of genuine education.
“础苍?AI programme might be pretty good at teaching you maths. It will teach you all sorts of technical skills, but I hope that we think that an education at a great university is also an important values conversation,” he said.
“I’m not sure it’s ever going to be a great education [with AI], which will always intrinsically be an interpersonal activity.”
Spence also argued that?AI will not only alter how universities teach, but also what they teach.
“We have to teach students to do the things the machines can’t do, how to use the machines, and also actually to do the same things as the machines - we didn’t give up teaching maths just because we got calculators.”
He added that universities would need to focus on the kinds of education that emphasised leadership, citizenship and values.
“That space of what you’re valued at educationally above what the machines can do, I think is going to be critical, and actually that’s what I think you’re going to be choosing your children’s education on.”
请先注册再继续
为何要注册?
- 注册是免费的,而且十分便捷
- 注册成功后,您每月可免费阅读3篇文章
- 订阅我们的邮件
已经注册或者是已订阅?




