Interest in artificial intelligence has ballooned, fuelling widespread expectations that AI technology could soon play a significant role across many industries. Talk of its benefits, however, is being matched by concerns over its environmental, social and governance (ESG) impact.
AI could have broad impact, from employment and content generation, to data security and privacy, energy consumption, and even diversity and inclusion. Beyond climate change, social equality, biodiversity and other more ‘traditional’ ESG issues, the emergence of generative AI in chatbots such as ChatGPT has sparked worries over far-reaching job robotisation, workforce displacement, and the potential for subversion and misuse, among other concerns.
This should prompt asset managers to rethink their internal ESG rating systems to incorporate potential AI risks and opportunities to ensure the risks of the technology do not undermine credible sustainability-related investment approaches and products, while taking advantage of related opportunities.
Putting a value on AI risks and opportunities
On the social side, AI may pose several ESG-related risks. Data privacy, AI bias and safety issues have been discussed by regulators, academics and industry players for a long time. Government and industry bodies have been issued initial guidelines such as the OECD/G20 AI Principles. However, few of these focus on social justice and long-term labour risks.
Estimates from McKinsey indicate that increases in productivity from generative AI that are likely to materialise when the technology is applied across knowledge worker activities could amount to USD 6.1-7.9 trillion annually, changing the anatomy of work as we know it and resulting in potential job displacements.
Indeed, in an interview with Bloomberg, the CEO of British Telecom estimated its headcount could fall by as much as 42% by 2030.
While AI might mean companies benefit from efficiency gains and increased profitability, they may need to reskill their workforce to use AI, while ensuring a fair and inclusive transition in adopting the technology. Before clear government policies are enacted on AI, companies are encouraged to conduct a human capital risk assessment, including the potential cost of reskilling or compensating unavoidable lay-offs.
Labour forces in sectors such as legal and professional services – tax and accounting, securities trading and brokerage – are likely to be among those most heavily impacted by AI, as well as business support services such as travel agencies.
Another challenge for investors will be to determine a company’s exposure to or use of falsified AI text and images and the potential financial implications, especially as long as there is little regulation.
On the environment side, AI poses several risks. According to a report in Harvard Business Review, the datacentre industry is currently responsible for 2-3% of global greenhouse gas emissions. It notes the volume of data across the world is expected to double in size every two years.
Storing data and refining AI models and algorithms through thousands of hours of training is data (and energy) intensive. Rising competition over AI among key countries and major market players will likely result in significant GHG emissions over the coming years.
Investors will need to assess the implications of AI use for their carbon footprint as well as the ways industry players are going to reduce emissions, for example by adopting efficient model and training techniques or heat recycling.
While the financial materiality of climate-related transition risk is becoming clearer, the ESG approach to AI risk is still virgin territory for many investors.
Incorporating AI risk in an ESG framework
At BNP Paribas Asset Management, we update our ESG rating framework periodically to consider the evolution of the global ESG context. Our model already considers AI-related risk such as privacy protection for the sectors most impacted such as IT and some financial institutions.
However, the possible wide applications of AI demand that it be further embedded into fundamental analysis across more sectors. For industries such as energy and utility, healthcare, professional services, media and advertisement, agriculture, environmental protection, it is worth thinking about how to incorporate AI into their ESG frameworks, both in terms of risks and benefits.
Within BNPP AM, we are seeking to do just so, all the while reviewing proposed AI regulation and industry research and considering how to engage with investee companies on AI-related risks.
This article is the first in a series of articles by tech analysts in the BNPP AM Sustainability Centre on AI and its implications for the sustainability agenda.