A leading UK law school has criticised students’ use of artificial intelligence (AI), arguing that the technology is often used to “cognitively offload tasks in a way that hinders learning”.
In a challenge to the sector’s embrace of generative AI use by students – as long as it is disclosed – scholars at UCL’s Faculty of Laws claim educators are selling their students short by allowing chatbot use, arguing that such “shortcuts to independent learning” are “tempting” but “undermine basic learning processes”.
A new position paper by UCL Laws’ dean and vice-deans explains that the faculty is using assessment which “guarantee[s] that AI does not substitute for the skills or knowledge acquisition being generated”.
“There are voices arguing that heavy use of AI in, for example, coursework, is acceptable – these tools will be available in the workplace, so why should students not have access to them in a university? Perhaps they can be used in assessment, as long as they are acknowledged? We disagree,” states the paper, published on 6 May.
Campus spotlight guide: Bringing GenAI into the university classroom
Speaking to Times Higher Education, one of the authors, Michael Veale, associate professor in digital rights and regulation at UCL, said encouraging students to use AI did not serve their interests.
“We’re not convinced that ‘using AI well’ is a skill that can just be taught in a module. In fact, the skills you need to use AI well are foundational skills that are hard to acquire – not something you gain by simply using AI,” said Veale, a machine learning expert.
“You’ll actually find the learning of skills is inhibited by just leaning into AI use,” he said, noting how tasks like summarisation, transcription and ideas generation can potentially be outsourced to AI but are useful parts of legal education and practice.
Relying on AI tools for “shortcuts” deprived students of moments that are crucial to building their confidence as independent free-thinking lawyers, continued Veale.
“When you talk to alumni, the things that they remember about their degrees are moments when they were scared, when they didn’t know the answers – moot courts or legal advice clinics where they realised they had the skills and knowledge to pull them through,” he said.
“If you have AI acting as the training wheels that you never take off, the bot whispering in your ear, you lose those transformative moments,” Veale added.
Making the pedagogic case for resisting AI use – and designing AI-proof assessment such as in-person exams and oral assessments – was more effective than simply attempting to ban its use on the grounds that it is a threat to academic integrity, said Veale.
“We haven’t played the game of detection or tried to fool ourselves into thinking we could rely on an insecure assessment framework involving coursework. Instead we’ve focused on pedagogic reasons [for resisting AI use],” he said.
Asked whether other universities and departments might adopt similar policies, Veale believed there is a “lack of vision around grappling with the ambient existence of AI that is not good for student learning”.
For instance, academics should push back on the idea that AI-written copy is acceptable in assessment, given they are supposed to assess “underlying abilities and skills rather than a capacity for ‘content creation’”, says the paper, which questions whether AI will have the transformational impact on the legal profession that many predict.
“Personally, I don’t think anything hinges absolutely on AI – its adoption is slowing down and businesses are floundering to use it,” said Veale.
That ambivalence about AI is also reflected in the paper, which calls on universities to “come together to respond to the future, double down on [their] values [and] cease passively responding to the ongoing evolution of business models of AI and cloud companies”.
jack.grove
