Portfolio Changes in the AI Era: How to Decide

Do you know the "yes, and…" rule?

This concept from improvisational comedy essentially means that to improvise successfully, scene participants must learn to welcome premises introduced by their partners (yes!) and then build on them (and…). It's a call to be flexible and willing to adjust to situations as they evolve. Doing the opposite—fighting premises and trying to force a scene into a new direction—risks killing the entire enterprise.

Investors could benefit from the same discipline. After all, they're always confronting new and potentially unsettling developments, whether in the markets, the economy, or in their own financial lives. How they respond could determine whether they successfully integrate and build on new facts as they come, or whether an unexpected twist disrupts their plans altogether.

Let's use the emerging and fast-moving world of artificial intelligence as a test case. Yes, AI presents legitimate risks. Companies are spending enormous amounts on AI, with uncertain returns. Data centers require financing, power, and public support. Regulation could slow adoption. AI services may eventually become commodities.

These matters are just for starters. What else could happen? How likely is each outcome? What expectations are already reflected in markets? Most importantly, does any of it warrant a portfolio change?

"Yes, and…" investors looking for an analytical framework to help them improvise through the AI era could use this "possibility, probability, price, and portfolio" framework when considering potential portfolio changes in these unsettling times.

Investors are wired to notice danger.

Research by behavioral psychologists Daniel Kahneman and Amos Tversky found that losses can have a far greater psychological impact than gains, which can scramble our thinking when we're trying to make decisions in uncertain situations. It's hard to dispassionately weigh pros and cons when our brains are on high alert because we're confronting something new and past experience is no guide.

AI provides no shortage of possible negative scenarios for us to fret over: Businesses could struggle to earn an acceptable return on their AI investments. Infrastructure-development projects could bog down in delays. Financing costs could rise. The technology may improve more slowly than expected.

Each possibility deserves consideration. But only as one possibility among many—and not as a foregone conclusion.

Periods of intense technological innovation tend to broaden the realm of what's possible. The future state of the technology, its uses, and its ultimate economic value become obscure and hard to map. This breadth may confound our instincts about the eventual path forward and trigger our innate aversion to risk.

However, just because the far end of the path is clouded with fog doesn't mean it'll end in disaster.

The "yes, and…" investor understands that they should take risks seriously by acknowledging possibilities, but without allowing their natural risk aversion to become the entire analysis.

After identifying what could happen, the next step is to consider what might make a particular outcome more or less likely. In short, how probable is a given possibility?

A positive case might conclude that AI could become more integrated into everyday life, contributing to productivity and economic growth.

But then one might consider potential execution risks. Companies are investing heavily in AI infrastructure based on expectations for future demand. That spending could slow if businesses fail to see sufficient returns, data centers encounter permitting or power constraints, or regulation makes implementation more difficult.

Those developments could delay the economic benefits, and markets could react sharply if returns arrive later than expected. However, neither of these situations would disprove the technology's ultimate potential.

Regardless, investors should watch the evidence as it develops. That includes AI-related revenue, adoption outside the technology sector, productivity, profit margins, and the relationship between capital spending and cash flow.

Once an investor has decided on what they consider the most probable path for AI investments, it's time to consider pricing.

One might take a positive view of AI investments but then conclude that market prices already reflect considerable enthusiasm for that possibility.

The question becomes: Has that enthusiasm gone too far?

The scale of the buildout adds to the sensitivity. Business investment in data centers has exploded in 2026. No company is an island. Their capital spending supports revenue throughout the supply chain. If just one of the major builders, known as "hyperscalers," slowed its plans, the effects could touch semiconductor manufacturers, utilities, industrial companies, and emerging-market suppliers.

The market is not ignoring every risk, however.

Software stocks were among the first to reflect concerns about disruption, commoditization, and pricing pressure. Investors are doing a better job of distinguishing among companies creating AI, supplying its infrastructure, and applying it. That suggests the market is showing more discrimination than broad index performance might imply.

Suppose that after sorting the probabilities from the possibilities and weighing the available assets according to their prices, an investor concludes that AI expectations have become too optimistic. What should they consider changing?

No investment decision is made in isolation, of course. It isn't enough to say that "stocks feel too risky," without considering the investor, time horizon, existing portfolio, and available alternatives.

Temporarily moving to cash may reduce an investor's exposure to any AI-linked market volatility in the near term. But it would also introduce inflation risk, reinvestment risk, and the possibility of missing out on further appreciation. Retreating to cash could engender another difficult decision in turn: when to return.

For many long-term investors, a more proportionate response may include:

• Rebalancing investments that have grown beyond their intended weights.
• Reducing unintended concentration in individual companies or sectors.
• Confirming that the portfolio's risk level remains appropriate.
• Maintaining diversification across asset classes, regions and market segments.
• Leaving an appropriate allocation alone.

A period of heightened uncertainty can serve as an opportunity to review a portfolio allocation. It is not, by itself, a reason to abandon it.

The process remains the same for any investor, even when their target allocations differ from others'. Goals, time horizon, and risk capacity should drive those differences, not the latest headline.

In this article, we're using AI to see how a "yes, and…" investor could confront and adjust to new facts in an emerging environment, but the discipline applies whether we're talking about inflation, debt, geopolitics, recession, or another disruptive technology. Investors will still need to decide what is probable from a wide universe of mere possibilities, consider whether markets are setting prices appropriately for a given probability, and determine whether the portfolio should change in response.

When the range of potential outcomes widens, diversification can help limit a portfolio's dependence on any single company, sector, region, or forecast. Long-term allocations help investors participate if innovation delivers while limiting dependence on any single forecast.

"Yes, and…" doesn't dismiss uncertainty. Rather, it allows investors to stay nimble and keep thinking through it.