Is AI’s real advantage better predictions or better decisions?

Investors spend enormous amounts of time and energy trying to forecast the future, whether it is the direction of interest rates, inflation, economic growth or market valuations. Yet markets have a way of surprising even the most informed investors. While a portfolio built on a correct forecast can deliver impressive results, consistently predicting the future is another matter entirely. However, could the benefit of more data, more computing power and general advances in artificial intelligence (AI) be a game changer?

The forecasting trap

Most investment committee meetings eventually arrive at the same debate: rates, inflation, economic growth and the direction of markets. While these questions matter, they can create a false sense of certainty. The danger is not in making forecasts, but in becoming too dependent on them. When portfolios are positioned around a base case expected outcome, investors are effectively making a bet that the future will cooperate. Unfortunately, markets rarely feel obligated to do so.

Few predicted the Global Financial Crisis. Few anticipated the scale of pandemic-related disruptions. Few foresaw the speed of the inflation surge that followed the pandemic and the aggressive interest rate response from central banks. Forecasts are attempts to navigate an uncertain future. They are most valuable when they challenge our thinking, highlight potential risks and test assumptions. Forecasts should shape the discussion, not dictate the decision. AI may help investors make better forecasts, but its greater value lies in improving how forecasts are developed, challenged and used. The most significant opportunity may be to facilitate better quality decisions in the face of uncertainty.

Decision-making quality

Decision-making quality is the thoroughness of the thinking and process behind a decision, independent of whether the eventual outcome turns out well or badly. This distinction is particularly important in investing because good decisions can produce bad outcomes, and bad decisions can produce good outcomes, especially over short periods.

Consider two investment committees. The first makes a concentrated investment based on a strong conviction about how the future will unfold. The forecast proves correct, the position performs exceptionally well, and the decision is celebrated as a success. The second committee takes a different approach. It rigorously evaluates assumptions, considers alternative scenarios, evaluates potential risks, and chooses a more diversified path. Markets move in unforeseen ways and short-term results disappoint.

Looking at only the outcome, the first decision appears superior. But did the committee actually make the better decision? Not necessarily. Chance, timing and unpredictable events often play a larger role than we care to admit when it comes to outcomes.

From a governance perspective, it is more clear-cut. The process adopted by the second committee, assessing whether the portfolio could withstand a wide range of outcomes, demonstrated higher decision-making quality. Moreover, sound decision-making processes tend to be more repeatable than accurate predictions. That is why many of the world’s leading institutional investors spend less time trying to be right and more time ensuring they make informed, high-quality decisions, regardless of what the future brings.

How AI can help

AI has the potential to improve forecasting and decision-making quality in several important ways.

• Improve the information going into the decision. The most obvious is the sheer amount of information it can process. Historically, forecasts were constrained by what an investment team could reasonably read, analyze and model. AI can expand that field dramatically, incorporating market data, economic indicators, company results, earnings-call transcripts, news and other forms of information.
• More information is only part of the story. AI can also identify complex, unstructured and non-linear relationships that investors might overlook. And unlike a traditional forecasting exercise that can become focused on a particular point of view, an AI-supported process can continually reassess the evidence as latest information becomes available.
• Challenge the investment thesis. A powerful role of AI is its ability to function as a systematic devil’s advocate. Investment teams naturally build convictions around their views. AI can provide a disciplined counterweight by actively searching for evidence that challenges those views, surfacing information that may have been discounted, and questioning how the thesis could turn out to be wrong. As the latest information emerges, it can also highlight what has changed since the original view was formed.
• Surface behavioural biases. AI can function as a broader disciplined challenge to the reasoning behind an investment recommendation, helping identify how behavioural biases may influence the decision. By systematically surfacing alternative interpretations, counterarguments and overlooked evidence, AI can encourage investors to separate what they believe from why they believe it. The goal is not to eliminate human judgment, but to make that judgment more balanced and defensible.
• Bring institutional memory into the process. AI can help teams search for previous investment cases, research and decision records to see if there was a similar decision before, what the assumptions were then and how it played out. This turns organizational knowledge into something more usable at the point of a new decision being made.

AI has limitations too. Much of what it learns comes from relationships contained in existing information, while markets can change precisely because something happens that does not resemble the past. The greater opportunity for AI lies in strengthening forecasting discipline: drawing on a broader range of information, challenging assumptions more rigorously, updating views more quickly, and recognizing sooner when the world has moved on.

Improving decision-making discipline

Even with better tools and more information, uncertainty cannot be eliminated from investing. The payoff from AI is not simply better forecasts, but better-quality decisions by strengthening the human process that surrounds it: challenging the objective, interrogating the evidence, widening the alternatives, stress-testing uncertainty, exposing potential biases and questioning whether the rationale is truly defensible. These following six questions provide a practical way to challenge your thinking, evaluate your assumptions and bring greater discipline to the decisions that matter.

1. Objective: Are you clear about the objective? High-quality decisions start with clarity of purpose. If we are not clear about the outcome we are trying to achieve, it becomes difficult to judge the alternatives, understand the trade-offs or determine whether the decision was successful.
2. Information: Are you using the most relevant information available? High-quality decisions depend on having relevant information, not simply more information. The goal is to identify what genuinely matters to the decision, separate useful evidence from noise, and recognize where important gaps remain.
3. Alternatives: Have other courses of action been considered? Decisions rarely come from choosing whether to accept the first idea on the table. Good decision-making means deliberately considering diverse ways of achieving the objective before deciding which path offers the best balance of opportunity, risk and trade-offs.
4. Uncertainty: Have you considered what could happen rather than relying on a base case forecast? The future rarely unfolds exactly as expected. Rather than building a decision around a base case forecast, a range of plausible outcomes should be assessed to challenge whether the decision still makes sense if the future looks different from the base case.
5. Behavioural biases: Have you recognized the risk of potential behavioural biases? Even with clear objectives and valuable information, our judgment can still be influenced by how we interpret that information. We naturally bring prior experiences, existing beliefs and preferences to a decision.
6. Process: Is there a clear rationale for defending the decision if the outcome disappoints? Because outcomes are influenced by uncertainty, even a well-reasoned decision can produce a disappointing result. A strong process creates a clear record of why we made the decision, what was believed, what alternatives were considered and what risks were knowingly accepted.

Together, these questions will not guarantee a good outcome, but they can significantly improve the quality of the decision behind it and help ensure that it was thoughtful, disciplined and made for the right reasons. AI can play distinct roles across these questions. For some, it can actively help develop the answer by analyzing information, exploring alternatives or testing outcomes. For others, its value may be as a disciplined reminder to challenge our thinking and ensure an important consideration has not been overlooked.

Leveraging AI for better quality decisions

The objective is not AI replacing human judgment when making decisions. It is better human judgment, supported by AI. Markets contain too much uncertainty for investors to judge skill simply by whether a decision was right. A better approach tries to build a disciplined process for making quality decisions by challenging assumptions, surfacing contrary evidence, identifying biases, considering alternative scenarios and making the reasoning behind a decision more explicit. The real promise of AI may not necessarily be making investors materially better at predicting the future, but to allow them to arrive at better quality decisions irrespective of what the future holds.