World Economic Forum warns of AI risks and urges ‘responsible’ adoption

June 25, 2024 (Preqin News) – The World Economic Forum (WEF) and CPP Investments have warned against the hasty implementation of generative AI, saying that short-term gains from rapid adoption may have damaging effects in the long term.

The new WEF-CPP Investments report, Responsible AI Playbook for Investors, cites Amara’s Law, which says that we often overestimate the short-term impact of technological advancements and underestimate their long-term effects. When ChatGPT launched at the end of 2022, many companies felt pressured to rapidly deploy AI for efficiency gains and to keep up with current trends. The alternatives industry was no different with myriad reports published in the last year on subjects ranging from automation of operational tasks (AIMA) and driving value creation through enhanced efficiency (BCG).

Early adoption is not without risks. The unintended consequences of rolling out AI applications too quickly can include ‘cyber-attacks, unwanted biases, job disruptions and/or displacement, and data leaks or poisoning,’ according to the WEF-CPP white paper.

AI is the hottest investment sector by some distance. Since the start of this year, Preqin has tracked 2,387 AI deals in venture capital (VC) globally, 24.1% of the total number of VC deals recorded. Even considering the increasing application of the AI label to companies that wouldn’t have attracted the tag a few years ago, this is a massive number. However, it does track behind 2023’s total of 5,343 VC AI deals and 2022’s total of 6,280. One of the largest private equity deals this year so far is Thoma Bravo’s $5.3bn take-private of UK-headquartered Darktrace Holdings, a cybersecurity AI firm, in April.

The paper calls for the implementation of ‘responsible’ AI (RAI) which, besides mitigating risks such as unwanted bias and job disruption, could promote growth through anticipating and adapting to legal and regulatory changes, improving customer engagement and retention by increasing trust, offering an advantage in procurement processes, and outperforming AI systems that are not ‘responsible’.

A definition of RAI is difficult to pinpoint and differs by industry, but according to the OECD, AI is considered ‘trustworthy’ when it can be ‘reliably developed and deployed without adverse consequences for individuals, groups, or broader society’.

For alternatives investors and other companies, there is a lack of authoritative guidance on AI metrics and disclosure on incorporating trustworthy or reliable AI. The European Commission, the OECD, and UNESCO have all published guidelines, principles, or recommendations on the ethical use of AI, but companies receive little practical guidance on how to practically implement these (although the OECD’s AI Policy Observatory now catalogs tools and metrics for trustworthy AI).

Nevertheless, investors expect a lot from AI. A recent survey by Alvarez & Marsal found that 97% of respondents saw digital infrastructure as key to their value creation plans, with AI a major factor. McKinsey’s latest forecast on generative AI states it could add the equivalent of $2.6–4.4tn to the global economy.

‘The broader business community is already struggling with risks such as data breaches, privacy loss, job loss, ethical challenges, misinformation, and disinformation,’ wrote Cathy Li, Head, AI, Data, and Metaverse, WEF, and Judy Wade, Managing Director and Head, Strategy Execution and Relationship Management, CPP Investments. ‘To protect their portfolios’ stability both now and in the future, investors should tackle both these immediate issues as well as the deeper implications of AI. This starts with establishing strong governance frameworks and clear principles and practices to integrate RAI standards into all applications.’

_{The opinions and facts included within the above do not constitute investment advice. Professional advice should be sought before making any investment or other decisions. Preqin providing the information in this content accepts no liability for any decisions taken in relation to the above.}