How could the AI revolution affect ESG?

Stephanie Maier, Chief Sustainability Officer at GAM Investments, considers the vast array of potential opportunities that artificial intelligence (AI) could enable in the sustainable investing space, as well as some of the risks that will need to be mitigated.

13 December 2023

Years from now, 2023 may well be known as the year where AI and complex machine-learning algorithms brought about a permanent shift in whole sectors of the real economy.

The launch of ChatGPT by OpenAI brought advanced AI-powered chatbots into the mainstream, showing how this kind of intelligence can trawl through vast amounts of online data to produce compelling human-like responses to queries. ChatGPT registered as the fastest growing internet application in history¹, reaching 13 million daily users in January 2023.

The financial world had already taken notice. According to Pitchbook², AI start-ups raised USD 15.5 billion in venture capital (VC) funding in the first half of 2023 as tech giants such as Google and Microsoft race to create their own competing products³.

The economic potential of AI is staggering, with the possibility to automate and streamline existing processes and create entirely new business opportunities that were unfeasible or impossible before, from devices which can monitor personal health, to platforms able to match talent with employers at record pace. However, in the rush to embrace AI, responsible investors need to be more mindful than ever of the risks that exist and apply those considerations across their portfolios.

The AI opportunity

We think AI has the potential to drive revenue, improve efficiency and cut costs for corporations. Its use can support important business needs⁴ such as automating processes, gaining insight through data analysis, and engaging with customers and employees. It is believed that businesses that can harness these technologies will be able to conduct data and time intensive processes much quicker and be able to spot emerging trends in the market landscape backed by powerful AI analysis. AI is already helping investors make informed decisions, with investors increasingly using natural language processing to analyse information such as reports, earnings calls and online news at scale.

Research by Infosys⁵ has found that firms leveraging AI can increase enterprise profit by 38%, and that AI could add USD 14 trillion in gross value to corporations by 2035. Goldman Sachs has estimated⁶ that generative AI alone could raise global GDP by 7% by 2033. These figures make AI development, and companies that effectively leverage AI in their business models, an attractive investment, in our view.

AI can be an effective driver of change towards more sustainable business models. Quantitative and qualitative investment data are a key challenge for responsible investors, and by analysing diverse data sets – from employee satisfaction to supply chain emissions numbers – AI algorithms can provide a more comprehensive understanding of a company’s environmental impact, labour practices and business conduct. AI can also be used to improve climate models, optimise the design and material for renewable infrastructure or detect and predict methane leaks from pipelines. This could have game-changing consequences in areas where vast data gaps currently exist, such as measuring biodiversity impact by tracking deforestation in real time through satellite imagery. Improved and expanded datasets will have a profound effect on investing, as investors can better understand their ESG risks and impacts, with this transparency starting to influence how capital is allocated.

There is convincing evidence that AI will help businesses become more profitable, transparent, and green – all important factors for sustainable investment. However, investors should also consider the risks of relying on artificial intelligence.

The other side of the coin

This use of AI has the potential to transform certain jobs and sectors: from administration to architecture, AI could replace more than 300 million jobs⁷. Resultingly, labour disputes are inevitable; already AI is one of the primary causes of one⁸ of the largest and most disruptive Hollywood actors and writers strikes in a generation, which by some estimates, has cost the California economy USD 3 billion⁹.

On a more technological level, the workings of AI models can be opaque due to their “black box” nature. Given its advanced capacity to make complex calculations and use multiple online sources, this risk should not be underestimated, particularly when this opacity may generate ‘phantom’ facts with limited reference to a direct credible source or the sources may be infringing privacy or copyright laws.

In a similar vein, the data AI systems are trained on can create bias. Despite advances in machine learning, AI is only as good as its initial programming. Algorithms based on existing, for example, racial, gender or geographical biases present in human decisions, or trained on data sets where certain groups are over or under-represented can further embed discrimination. Amazon¹⁰ stopped using a machine-learning tool in its recruitment following the discovery that it was not gender-neutral, favouring male applicants over female ones. The black box nature of AI makes accountability, and overcoming these biases, challenging.

Investors also cannot ignore the fact that AI itself has a significant carbon footprint. Creating and deploying AI models requires a substantial amount of computing power. A University of Massachusetts study¹¹ found that simply creating a single AI model requires 626,000 lbs (approximately 284,000 kg) of CO2, the same as 62 cars driven continuously for a year. Given that there are approximately 14,700 AI-driven companies in the US alone¹² (a number that will surely grow), that amounts to a worrying amount of emissions over time. One study¹³ showed that global data centre electricity use was approximately 1.15% in 2016; despite efficiency gains this could double by 2030.

Risks to markets

AI is a tool that will inevitably be adopted by our industry and the industries in which we invest, and our goal is to use it appropriately, understand its limitations and engage constructively with structural challenges.

Given the potential pitfalls of AI, considering the following factors will be important:

Transparency – Have steps been taken to make the function of AI models (as well as data compilation) as available and transparent as possible? Have the developers made sure that the AI model is provided with data that minimises bias?

Applicability – Is the AI appropriate for the task? Does it provide accurate and useful data? Has it been tested for defects that could cause harm?

Rights and responsibility – Have safeguards been put in place to reduce potential harm to the business and broader society? Have privacy standards been upheld regarding the data used to build the model?

AI has shown incredible potential to transform investments, boost profits, increase the availability of critical data, and drive more sustainable outcomes. However, the structural impact and risks are likely to lead to increased regulation and enforcement. While the regulatory approach is still being developed – with different approaches in the US, EU and UK – the far-reaching impacts of this technology will continue to attract regulatory and investor scrutiny for both the opportunities and risks they bring.

For more thoughts on AI, read Mark Hawtin’s latest article: The Disruptive Strategist: The AI capex bubble – is history repeating itself?