Sustainable use of AI and the role of law firms

Australia’s largest law firms are certainly endeavouring to use AI to its full potential and have eagerly embraced such new technological capacities. This is, of course, partly borne out of necessity – the evolution of lawyers’ roles in the face of such tech, and clients’ apparent preference for AI for legal advice over that of human lawyers, means that this year is shaping up to be pivotal in defining the future direction of legal service delivery.

However, such usage of new and emerging technologies cannot come at the expense of law firms’ stated commitments to environmental, social, and governance (ESG) considerations.

Firms of all stripes – not just domestically but also abroad – have espoused the importance of sustainability, from meaningfully advising clients on the legal implications of greenwashing through to ensuring more energy-efficient office spaces. ESG has become a key trend for the practitioners (if not one of the most urgent), and a headline priority for business leaders, including because of its potential to drive M&A and capital markets activity.

Earlier this year, the Massachusetts Institute of Technology reported that the computational power required to train generative AI models that often have billions of parameters “can demand a staggering amount of electricity, which leads to increased carbon dioxide emissions and pressures on the electric grid”. Moreover, the education provider noted, “a great deal of water is needed to cool the hardware used for training, deploying, and fine-tuning generative AI models, which can strain municipal water supplies and disrupt local ecosystems”.

Elsa A. Olivetti, a professor in the Department of Materials Science and Engineering and the lead of the Decarbonisation Mission of MIT’s new Climate Project, said: “When we think about the environmental impact of generative AI, it is not just the electricity you consume when you plug the computer in. There are much broader consequences that go out to a system level and persist based on actions that we take.”

VIEW ALL

To this end, Lawyers Weekly approached numerous BigLaw firms across the country to ask how such national and global practices can uphold their sustainability commitments in an age of increased AI usage. Specifically, firms were asked about the environmental concerns around the elevated use of new technologies, including and especially AI, and how large firms can manage their usage sustainably, in accordance with pre-existing ESG and CSR commitments.

In short, can the mainstream use of AI in law firms be compatible with meeting ESG targets?

Lawyers Weekly posed this question to firms two years ago, but – with the benefit of more time to get to grips with the capacities, limitations, and dangers of such technologies – answers can and should be demanded of the biggest providers of legal services nationwide.

Allens, Baker McKenzie, King & Wood Mallesons, MinterEllison, and Piper Alderman all declined to comment or did not meet Lawyers Weekly’s publication deadline.

Dentons partners Michelle Segaert and Vanessa Gore, told Lawyers Weekly the “well researched and reported on” energy usage of AI means that law firms “need to start thinking about the energy consumption issues of housing their own AI solutions in their own data centres, increasing their scope 1 and scope 2 emissions compared to buying an outsourced AI solution increasing their scope 3 emissions (and potentially scope 2 emissions)”.

In addition, the pair said, “AI policies need to consider what is an efficient usage of AI in the law firm environment having regards to the energy usage of AI, in addition to the ethical and legal issues”.

Ben Lehman, who is the head of digital transformation at TG Legal + Technology and Thomson Geer, reflected that the world is still at the very early stages of its development when it comes to resource-intensive GenAI.

“Even cutting-edge research companies have trouble forecasting compute requirements,” he said.

That does not mean, Lehman noted, that the potential impact of something that will become commonplace should be ignored.

“The net environmental impact of AI is still very uncertain. In recent months there has been significant improvement in AI model efficiency. Off-the-shelf, open-source AI models are now available, providing practical competition to much larger, more resource-intensive proprietary models,” he said.

“For many legal workflows, which require a human lawyer in the loop, these smaller, more efficient models can provide a better cost and environmental balance compared with using the biggest and best AI just because you can.”

Successful firms, Lehman continued, will match the right AI to the right job for an overall better outcome.

“[It is] also worth considering is that the legal industry largely deals with documents and typically a lot less images and video, both of which consume significantly more processing resources. AI use involves trading something done manually for something done more efficiently by AI, rather than adding something entirely new (and therefore adding new resource requirements to the work product),” he said.

Elsewhere, he said, the potential sustainability impacts of AI may well be significantly different by industry or even by task.

“As AI use becomes more targeted, and compute demands even out, the more thoughtful and nimble firms will be able to balance this clarity with existing sustainability commitments and client service,” Lehman said.

Segaert and Gore said that law firms like Dentons are “just starting to grapple with the sustainability issues of using AI”.

“Many regions within Dentons globally have been undertaking boundary setting and measurements for scope 1, scope 2, and scope 3 emissions over the past couple of years (including Australia) and in Australia, we are now in the process of refining our measurements for our scope 3 emissions,” they said.

“With Dentons UK being a developer of an in-house AI product – Fleet AI – along with making decisions as to which third-party AI service providers to partner with, we are starting to consider the scope 1, scope 2, and scope 3 emissions impacts of developing and housing AI solutions as well as buying (outsourcing) AI solutions.”