Banning the use of artificial intelligence (AI) in higher education risks cutting off discussions about how to innovate pedagogy, according to learning experts at Google Deepmind.
Miriam Schneider, director of learning initiatives in the company’s main AI research division, said AI doesn’t have to “change pedagogy” but rather can “enhance it.”
As many universities grapple with how best to approach the use of AI in assessment and learning, she said the “conflict” between “what should remain human” and what can be done with AI “could play an important role”. Times Higher Education.
“This basically gives us an opportunity to reflect and allow us to address some of the big questions that we had almost vaguely assumed,” she said.
“I think it will accelerate the conversation about what school system design looks like to best incorporate more holistic learning for students, rather than just the transfer of knowledge,” said Schneider, who has worked at Google for more than 20 years. Last September, she moved to DeepMind, the company’s main AI research institute, which funds a variety of scholarships and university research.
“I think those are important conversations,” she said. “I think it’s all or nothing, and there’s a risk that we won’t get them in the end.”
Conversations are increasingly focused on “what the future models of learning and teaching will look like as technology advances further,” Schneider said, adding, “Interestingly, it has less to do with the technology itself and is instead thinking about the role between teachers and students.”
“How do you center humanity? How do you center connections and relationships? How do you motivate people? In some places, technology is good at that, and in others, it’s not.”
Following criticism that large-scale language models (LLMs) are being misused by students to cheat in academic environments, tech companies have launched various “learning modes” to combat this. Google’s version of Gemini, “Guided Learning” mode, leverages the LearnLM methodology that powers Google’s education products, including Google Classroom.
talk to The At last month’s World Education Forum, Schneider said the challenge for education technology manufacturers is “how do we bring in outsiders” to provide “information” on how AI is optimized for learning.
But to do that, he says, “we need to treat learning as a science. That is, there are core fundamental truths in the science of learning, and we need to build technology on top of them, not the other way around.”
Expecting LLMs to be “good at teaching and learning” without prior training is “like taking a stranger off the street, putting him in front of your classroom and saying, ‘Wow, now you’re a teacher.'”
“To strengthen what we know about the learning sciences, we need to purposefully structure LLMs to be fundamentally tuned to excel in pedagogy, and not assume that that will happen without some intervention.”
