91ÇÑ×Ó

A rush to embrace artificial intelligence risks hampering universities¡¯ net zero targets, with leaders being accused of generally overlooking the ¡°hidden¡± huge environmental impact of the new technologies.

Universities are increasingly using AI in every aspect of their operations, rolling out chatbots and AI assistants to help in areas such as teaching, admissions and well-being support.

This has been accompanied by a surge in staff and students using large language models such as ChatGPT in their research and assessment.

Alex de Vries, founder of Digiconomist, which examines the impact of technology trends on the environment, said that rising energy use was an inevitable consequence of such ambitions.

¡°This is a very energy-intensive technology that just drives up power demand. And if you drive our power demand, you¡¯re going to end up increasing your carbon emissions and water consumption¡±, he told?Times Higher Education.

Across the globe, big technology firms have seen their carbon emissions rocket in recent years following the advent of AI, which is far more energy-intensive than traditional forms of technology. In its??released last year, Google revealed that its carbon emissions had increased 48 per cent in five years, primarily driven by increases in its data centre energy consumption and supply chain emissions.

Speaking at a recent event at the London School of Economics, Ravi Pendse, vice-president for information technology and chief information officer at the University of Michigan, admitted that its ambition to be ¡°the first university in the world to provide a custom suite of generative AI tools¡± was in ¡°direct conflict with our environmental goals¡±.

The university was trying to get to grips with the issue, Pendse said, by establishing working groups and has included add-ons on its AI tools to tell students how much electricity it has used.

A challenge for universities trying to track the environmental impact of their AI use was a lack of transparency from the likes of Google and Microsoft, said de Vries, also a PhD candidate at Vrije Universiteit Amsterdam studying the environmental impact of technologies.

Any emissions generated using technologies provided to these companies is effectively outsourced, blurring transparency and accountability, he said.

As a result, according to de Vries, there was a lack of thought going into how such tools are being rolled out across universities as ¡°no one really knows what¡¯s going on¡±.

¡°We don¡¯t talk about the environmental impact of AI enough,¡± agreed Sophie Rutschmann, faculty lead for digital education in immunology at Imperial College London. ¡°I think the message at the moment is that there¡¯s an environmental impact, and that¡¯s it. We¡¯re stopping there.¡±

Campus spotlight guide: A greener future for higher education

Instead, universities and academics ¡°should push it and should engage critically with that environmental impact so that individuals can make the right choices, [and] teach students that they can use AI when it¡¯s actually going to enhance their work, versus just using it for gimmicks and because it¡¯s there¡±.?

Rutschmann said that AI has the ability to ¡°bring a revolution to how we do healthcare¡±, so the emphasis should be on explaining to students, ¡®¡°OK, this is a smart use of AI because it¡¯s really advancing research¡¯ and providing examples of the good use of AI¡± so that it is only used for when it can really add value.?

It was ¡°futile¡± and ¡°too late¡± to tell students to cease using AI, she said,?pointing out that some Russell Group universities now allow the use of AI in their assessments.

De Vries said universities and individual users of these technologies ¡°can¡¯t necessarily be blamed¡± if big tech firms are not being transparent about the impact of these technologies, but said universities need to pressure?the firms into making their environmental data more transparent to make AI use more responsible.

¡°It¡¯s just fundamentally very hard to combine this technology with environmental sustainability simply because from an AI perspective, bigger is better¡­but from an environmental perspective, the ¡®bigger is better¡¯ mentality is a disaster,¡± he said.

juliette.rowsell@timeshighereducation.com