FRESH TAKES ON BIG IDEAS

In private equity, much of the conversation about AI has centered on what the technology can do: the tools investors should use, the data that should power them, the capabilities firms should build, and even the role of AI in investment committees.

These topics matter. But I keep coming back to a different question: How can we redesign the investment process for a world where AI can produce a good first-pass answer almost instantly, when the judgment required to turn that answer into conviction, action, and ultimately better investment outcomes remains scarce?

It’s an odd paradox. AI can radically increase productivity in investing, stripping out mechanical work and generating more data, more scenarios, and more analysis in a fraction of the time. That’s real value when private equity firms are deciding whether to deploy or harvest capital is the prudent, risk-adjusted choice. But more analysis does not automatically produce better judgment. There are still only 24 hours in a day, and human attention doesn’t expand just because output does. AI may multiply the analysis available to investors, but it cannot multiply their capacity to absorb it. 

Moreover, if AI takes over work that once built judgment and expertise along the way, we cannot assume that the next generation of private equity leaders will continue to develop those capabilities on their own. Recent research offers an early warning. A 2026 study tracking 26,811 Chinese secondary students over 30 months found that using generative AI helped them finish homework 30 percent faster and that their homework scores rose 18 percent. Within six months, however, the same students’ closed-book exam scores had fallen 20 percent. The decline was concentrated among students who appeared to be outsourcing more of the work to AI.

A related study from Wharton indicates that AI usage can also affect the quality of human decision-making. Researchers found that accurate AI advice substantially improved people’s accuracy, but when AI confidently gave the wrong answer, people who used it performed worse than those who did not use AI at all. This is termed “cognitive surrender”: People adopted AI’s confidence at the expense of their own judgment. 

“If AI takes over work that once built judgment and expertise along the way, we cannot assume the next generation of private equity leaders will continue to develop those capabilities on their own.”

These findings have implications for investing. The objective of diligence has never really been to produce an answer. Seasoned investors know from the start that the investment model will be wrong—I’ve never built one that wasn’t. The work lies in understanding by how much, in which direction, and why. Some of that is knowable through analysis. Ultimately, though, every investment rests on a judgment about what is likely to happen next, and no amount of AI-generated analysis can make that judgment for us. Early in my career, when I was building investment models, the real learning rarely came from typing in formulas or gathering data. It came from living inside the model: changing an assumption, getting an unexpected result, and working out whether I’d made an error or stumbled onto something real about the business. 

The opportunity for private equity firms, then, is to redesign the investment process so that AI-generated analysis strengthens human judgment. In practice, this means making sure investors form views before seeing the machine’s answers, identify what evidence could change those views, interrogate surprising outputs rather than accept them, and explain investments in language that demonstrates real understanding.

I recently heard a diligence process half-jokingly described in terms that capture this shift: two days to get an answer out of AI, three days to figure out what the answer meant, and a week to work out how to communicate it in a way that could build conviction within the investment committee. 

The private equity investment process is full of such moments in which human judgment needs to be formed, exercised, and tested. Before diligence even begins, someone has to decide how best to spend that time. Set the bar one way, and we risk passing on good deals; set it another, and we spend time on companies that were never going to clear it. How we balance those risks is itself a judgment and a design choice worth making deliberately.

That is why I don’t think the firms that thrive in the AI age will necessarily be those with the best models, the most data, or the most agents. They’ll be the ones that keep building judgment in the next generation of investors while ensuring that the judgment accumulated over decades around the investment committee table doesn’t quietly erode.

—Edited by Arshiya Khullar, senior editor, Gurugram

ABOUT THIS AUTHOR

Andrew Mullin is a senior partner in McKinsey’s Toronto office.