02 September 2026
Artificial intelligence promises to make investment analysis faster, broader and more sophisticated. But what happens when greater analytical efficiency comes at the expense of the very capabilities investment professionals need to make sound decisions?
A recent Enterprising Investor analysis explores this tension through the concept of cognitive delegation: the growing tendency to rely on AI-generated analysis before developing an independent understanding of the problem being examined. The risk is not simply that AI may produce incorrect answers. More fundamentally, excessive reliance on these systems could gradually weaken investors’ ability to question assumptions, defend investment theses and adapt their reasoning when market conditions change.
Investment expertise has traditionally been built through a demanding process. Analysts gather information, construct models, test assumptions and confront evidence that may challenge their initial conclusions. These activities are time-consuming and sometimes inefficient, but they also develop the tacit knowledge and mental frameworks required to exercise judgment under uncertainty.
AI can dramatically reduce that friction. An analyst can now produce sophisticated financial models, investment theses or risk assessments in a fraction of the time previously required. The efficiency gain is significant, but it creates a potential paradox: professionals may become capable of producing increasingly sophisticated outputs without developing an equally sophisticated understanding of how those outputs were reached.
This issue is particularly relevant for the next generation of investment professionals. Junior analysts have historically developed expertise by progressing through increasingly complex analytical tasks. If those tasks are delegated too early to AI, firms may inadvertently remove part of the learning process through which investment judgment is formed.
The consequences may only become visible when conditions depart from the model. An investment thesis can appear convincing on paper while concealing weaknesses in the analyst’s underlying understanding. In live discussions or periods of market stress, however, professionals must be able to defend assumptions, respond to counterarguments and revise their conclusions as new information emerges. Those capabilities cannot simply be outsourced.
Nor is cognitive delegation exclusively a junior-level problem. As AI-assisted analysis becomes routine, experienced professionals may also become less accustomed to maintaining the internal mental models they use to interpret complex market environments. This can become particularly problematic precisely when independent reasoning matters most: during high-stakes decisions made under significant time pressure.
Who is accountable when AI informs the decision?
The second challenge identified by the analysis concerns accountability.
As human judgment and machine-generated inference become increasingly intertwined, determining where responsibility lies becomes more complicated. Yet regulatory, legal and ethical frameworks ultimately hold people and organisations - not algorithms! - responsible for investment outcomes.
For asset managers, this makes AI governance an integral part of the investment process rather than simply a technology issue. Firms need clearly defined decision rights, transparent documentation and audit mechanisms capable of establishing how AI-generated analysis contributed to a final investment decision.
Without those safeguards, firms risk developing what the authors describe as a “black-box dependency”: investment conviction based on analysis that professionals cannot fully understand or effectively challenge. The consequences can extend beyond investment performance to regulatory, reputational and client risks.
AI also raises a less obvious organisational question: how should investment skill be measured?
Performance attribution has traditionally helped investment firms identify individual expertise, allocate compensation and determine career progression. As AI becomes deeply integrated into research and portfolio management, separating human contribution from machine contribution becomes increasingly difficult. If firms cannot determine whether superior results reflect genuine investment judgment or effective reliance on AI tools, evaluating talent - and rewarding it appropriately - becomes considerably more complex.
AI as an amplifier, not a substitute
The analysis ultimately points towards a different way of thinking about AI adoption. The objective should not be to eliminate human involvement from investment processes, but to use technology to amplify the capabilities that remain distinctly valuable.
As access to data and computational power becomes increasingly widespread, those resources alone may provide less differentiation. Judgment and accountability, by contrast, could become scarcer - and therefore more valuable.
This suggests that successful investment organisations may deliberately preserve a degree of “constructive friction” within their processes: encouraging professionals to develop their own views before consulting AI, challenging machine-generated conclusions and ensuring that someone remains clearly responsible for the final decision.
For investment professionals, AI therefore presents a challenge that goes beyond learning how to use increasingly sophisticated tools. The more capable those tools become, the more important it may be to preserve the ability to reason independently of them.
In an investment industry built around decisions under uncertainty, the competitive advantage of the AI era may ultimately belong not to those who delegate the most thinking to machines, but to those who understand which thinking should never be delegated at all.