Ali Mesbahian is an IPilogue Writer and a 2L JD Candidate at Osgoode Hall Law School.
On June 28, 2021, IP Osgoode hosted a panel discussion in their Bracing for Impact Webinar Series titled AI’s Dirty Footprint. Organized in collaboration with the Harry Radzyner Law School at IDC Herzliya in Israel, Microsoft Canada, and Alectra’s GRE&T Centre, the central question of this webinar was: in what way can we use artificial intelligence (AI) to ensure that the negative impacts of its energy consumption do not exceed its beneficial effects for environmental sustainability?
IP Osgoode’s own Professor Giuseppina (Pina) D’Agostino opened the panel by setting the stage for the discussion and introducing the speakers.
In his opening remarks, Dr. Amir Asif, Vice President of Research and Innovation at York University, noted that AI remains a “key strategic area” for research at York. Emphasizing the need for an interdisciplinary approach, Dr. Asif also stated that exploring AI’s ethical and legal implications will require collaboration between researchers in the AI community, social sciences, and the humanities.
Indeed, “collaboration” was one common thread among all the speakers. In his first formal address since he took office, the Hon. David Piccini—Ontario Minister of the Environment, Conservation and Parks—turned not only to universities for ideas in using AI to improve the climate situation, but also to youth as part of the government’s broader environmental policy plan. The youth, he noted, must be “unapologetically engaged”. Given the Ford government’s attempt to strike down a Youth-led climate action that challenged the constitutionality of Ontario’s plan to reduce greenhouse gas emissions to only 30 percent below 2005 levels, the Minister’s words are welcome if they signal any change.
The panel then proceeded to a discussion moderated by Dr. Aviv Gaon, professor at IDC Herzliya. In his introduction, Gaon brought attention to a research article published by Springer Nature Limited that outlines 17 internationally-agreed-upon sustainable development goals for 2030, spread across 169 targets. With respect to the environment, the study shows that AI’s potential to enable the environmental targets of these 17 goals outweighs its inhibitory effects.
The first panelist was Andrea Roszell, Director of Energy, Sustainability and Infrastructure at Guidehouse. Her discussion was centred on AI’s capabilities to increase efficiency in the energy and utility sector. In particular, she pointed to the “energy cloud”, a concept developed at Guidehouse that moves away from a “one-way flow” of power from energy centers to consumers, to a more networked, interconnected “multi-flow” dynamic. This requires an infrastructure—a neural grid—that utilizes artificial intelligence in technology, such as sensors software and monitoring systems, to create large “data sets” for utilities to access. Despite requiring increased energy consumption, Roszell stated that these data sets are a net benefit to the environment due to the new efficiency gained in management of greenhouse gases and predictive maintenance models that ultimately lead to a more sustainable and reliant energy infrastructure.
The second panelist was Dr. Audrey Lee, Senior Director of Energy Strategy at Microsoft. She started by pointing to Microsoft’s 2020 Environmental Sustainability Report. Among other goals, Dr. Lee highlighted Microsoft’s plan to offset all of its electricity usage with renewable energy by 2025 and to be carbon negative by 2050. Lee noted, however, that the first step in achieving any such goal is to establish a proficient “measurement infrastructure” that can enable us to quantify our environmental footprint with sufficient precision—for example, data analytics that detail how and to what extent a particular utility uses electricity at each hour.
The panel then continued to its third speaker, Kapil Singhal, Co-Founder & CEO of Vyntelligence. At the very outset of his discussion, he too emphasized the need for collaboration. In particular, Singhal noted how Vyntelligence has made possible a new form collaboration between artificial intelligence and human brain power. Utilizing short videos of workflow in the field, artificial intelligence can augment workers’ awareness of a given project by revealing further areas of risk and benefit. This, when combined with human cognitive and decision-making power (which Singhal noted far exceeds what AI can learn), will yield more efficient outcomes. One such outcome is enhancing the infrastructure that allows for remote work (the importance of which is vividly felt in times of COVID-19), reducing thereby the carbon footprint of work-related travel.
Finally, the panel featured Neetika Sathe, Vice President of the GRE&T Centre at Alectra Inc. First, she noted that as more and more people gain access to the internet, global energy consumption is bound to increase. Thus, she emphasized the need for international collaboration beyond local efforts. She further mentioned that about half of the energy used at datacentres is used to cool their servers, which brings attention to the need for more efficient infrastructures.
In closing, it is important to address that, as the panelists mentioned, data centres account for only 1-2% of global energy consumption. However, as I mentioned in my earlier article, AI’s “dirty footprint” is not confined to the energy it consumes, but extends to its ability to offer services for resource extraction which, for example, is enabled by the connection and collaboration between the tech and fossil fuel industries. Any meaningful policy directed at reducing AI’s negative environmental impacts must also account for this broader perspective.
A link to watch a recording of the event can be found on IP Osgoode’s Past Events page.