Roundtable: Will AI transform private equity?
A feverish excitement has developed around advances in artificial intelligence over the past 18 months, but how will it affect private equity and venture capital investing? Will it be used at the portfolio company level to help with decision-making? Or perhaps by limited partners to improve fund selection processes? Three research papers explore these areas, and we discuss the findings with the authors and three seasoned practitioners.
PE and VC firms – along with their LPs – are experimenting with AI-driven data analytics technologies, including machine learning and natural language processing, as they seek to enhance decision-making processes and improve selection techniques.
The efficacy of AI in a private markets context remains largely unproven, however, and many remain sceptical of the claimed benefits. Will algorithms really replace intuition in the relationship-driven realm of fund selection? Can a technology trained on historical information predict breakthrough innovations for VCs? Can AI help build better boards?
Three authors of academic research on AI in VC, in fund investment selection, and in company board appointments explain their findings and discuss them with three practitioners.
Chaired by Amy Carroll
Meet the speakers
![]() |
Maxim Bonelli
Maxime completed his first PhD in mathematics at the National Institute for Research in Digital Science and Technology. He subsequently worked as a quantitative researcher in the asset management industry, before returning to academia to complete a second PhD, in finance, at HEC Paris.
|
![]() |
Anne Glover
Anne worked in manufacturing and strategy consulting before joining Apax Partners in 1989. She became COO of one of her investee companies, Virtuality Group, after it listed in London, and then returned to investing as a business angel before founding Amadeus with Hermann Hauser in 1997. She is currently serving a term as non-executive director in the Bank of England’s Court of Directors.
|
![]() |
Merrick McKay
Merrick joined abrdn in 2014. He is responsible for the firm’s European PE business. Before joining abrdn, he was head of European investments in Macquarie’s private markets division. He was also a partner at Primary Capital.
|
![]() |
Léa Stern
Léa is an assistant professor in the finance and business economics department at the University of Washington, Foster School of Business. She is also a visiting scholar at the Allen Institute for Artificial Intelligence.
|
![]() |
Ludovic Phalippou
Ludovic specialises in PE and asset management research. He has a degree in economics from Toulouse School of Economics, a master’s in economics and in mathematical finance from the University of Southern California, and a PhD in finance from INSEAD.
|
![]() |
Joshua Lowe
Joshua joined Coller Capital in 2020, where he is responsible for the investment team’s application of data science. He has a decade of experience in applying machine learning to solve commercial challenges and was previously a lead data scientist |
Limited Partners versus Unlimited Machines; Artificial Intelligence and the Performance of Private Equity Funds, by Borja Fernández Tamayo and Flo...
The research finds that quantitative information, such as track record, is correlated to the success of the fundraising but not to the performance of the fund. However, natural language processing tools are able to use qualitative information contained in the strategy section of prospectuses to predict fund performance with a meaningful degree of accuracy. The technique accurately classified 67% of underperformers (funds in the bottom third of performance) and 75% of outperformers (funds in the top third of performance), supporting the view that investors should be taking qualitative information into greater account in their asset manager selection process.
Selecting Directors Using Machine Learning, by Isil Erel, Chenhao and Michael Weisbach (both Ohio State University) and Léa Stern (University of Washington, Foster School of Business), Chenhao Tan (University of Colorado), explores a novel alternative to the existing practices of nominating committees, which often rely heavily on personal networks. Instead, the research uses algorithms based on company data, current board member information and the attributes of potential directors to predict the performance of directors being considered for a board position.
The algorithms were trained on a set of directors appointed to public companies between 2000 and 2011. The directors that the algorithms predicted would do poorly did worse on average than those predicted to do well, with success and failure measured both by shareholder approval and short tenures. The research also found that male candidates with large networks and large directorship portfolios were the most likely to underperform.
The results have significant real-world implications for the corporate governance that surrounds the critical process of appointing board directors, and could prove important when tackling the challenge of enhancing diversity at a board level.
The final piece of research, Data-driven Investors, by Maxime Bonelli (London Business School), examines the role of AI in making VC investment decisions. The research is predicated on the idea that data technologies are inevitably designed to identify historical patterns, while VC firms focus on identifying novel ideas.
Using Crunchbase data on global investments, acquisitions and IPOs of start-ups, as well as information on patent applications and grants, it finds that VC firms that have adopted data technologies are more likely to invest in companies similar to those that they have invested in previously, and that they become better at avoiding failures within this pool. However, those VC firms are also less likely to identify rare major successes – the coveted VC home run.