A few months ago, I read Coming into View: How AI and Other Megatrends Will Shape Your Investments by Joe Davis, Vanguard’s Chief Economist.

It’s a thoughtful book that I recommend. It isn’t overly technical, and it frames the biggest questions investors face today: AI, debt and deficits, demographics, potential deglobalization, and—hardest of all to measure—geopolitical risk.

The core question Davis keeps returning to is deceptively simple: can AI generate enough productivity growth to offset the growing debt burden?

Although he presents it as a close race, Davis ultimately leans optimistic. He thinks AI will be transformative enough to forestall the debt problem.

What I like most about the book is the way Davis frames outcomes in explicit probabilities, rather than single-point forecasts. His base case—what he calls the “AI Transforms” scenario—carries roughly a 45 to 55 percent probability, making it the most likely outcome, but far from a certainty.

Here’s what that scenario looks like in his estimates:

• GDP growth of 2.5 to 3.5 percent
• Inflation between 1.5 and 2.0 percent
• Bond yields between 3.2 and 4.2 percent
• Stock returns between 6.0 and 13.5 percent

That sounds pretty good. The ranges are wide, but that’s actually part of the appeal—precision would be less believable here. The part I like less is that, stripped down, it’s still just a coin flip with some statistical confidence bands around it.

The darker scenario—which Davis assigns a 30 to 40 percent probability—is more sobering. In this version of the future, AI turns out to be fascinating and widely used, but economically disappointing. It doesn’t meaningfully boost profits or productivity.

Think social media: enormously impactful culturally, extremely profitable for a handful of companies, but economically neutral—or even destructive—for others. We all use it, but it didn’t raise the economy’s speed limit.

If AI fails to deliver productivity gains, the debt overhang becomes binding. In that world:

• GDP growth slows to 0.5 to 1.5 percent
• Inflation runs higher at 2.0 to 3.4 percent
• Bond yields rise sharply to 5.9 to 7.2 percent—levels that rival today’s emerging markets
• Stock returns land between –1.3 and 6.1 percent

Those returns look especially poor once you adjust for inflation.

One gap in Davis’s analysis is unemployment. While he is generally constructive on labor markets, I wish he had provided specific unemployment estimates alongside his growth and inflation assumptions.

The third scenario, which he assigns a 15 to 20 percent probability, is essentially a continuation of the status quo: about two percent growth and two percent inflation. Davis describes it as living in 2017, but with video calls that actually start on time.

Davis spends substantial time discussing how to position portfolios under each of these regimes. I’ll pick up that thread next week.

The takeaway for now is that AI offers a credible solution to a problem that has festered for decades and is increasingly out of control under both political parties.

But much remains uncertain. I use AI every day, and while it’s interesting and useful, it hasn’t yet been transformative for me. I regularly encounter factual errors and can easily steer queries to produce almost any answer I want, which underscores how immature the technology remains beneath its sophisticated veneer.