Measuring a Transformation as It Happens
One of the deepest problems in understanding AI's effect on the economy is that our instruments for measuring it are too slow. Traditional economic statistics arrive with lags of months or years, by which point the changes they describe are already history. Stanford's Digital Economy Lab, part of the university's Institute for Human-Centered AI, has launched a response to that problem: a freely accessible platform called the AI Economic Indicators, designed to track AI's impact on work in something much closer to real time. The premise is that a transformation moving this fast demands measurement that moves fast too, and that the conventional data infrastructure was simply not built for it.
Susan Young, Director of Strategic Initiatives at the Stanford Digital Economy Lab, articulated the gap the project aims to fill, noting that one challenge with major technological change is that it can take years for traditional data sources to fully capture what is changing. That lag is not a minor inconvenience. Policymakers, educators, and business leaders are making consequential decisions right now about how to prepare for AI's labor effects, and they are largely flying blind, working from data that predates the technology's most rapid advances. A real-time measurement system, even an imperfect one, is a meaningful improvement over waiting for the official statistics to catch up.
Three Windows Into the Change
The platform launched with three distinct dashboards, each offering a different vantage on how AI is reshaping the economy. The Canaries Dashboard, built in collaboration with ADP Research, tracks employment trends for occupations with varying degrees of AI exposure, with particular attention to early-career workers aged 22 to 25, who make up a defined slice of the sample baseline. The name is apt. The premise is that certain jobs function as canaries in the coal mine, showing the effects of AI earlier and more sharply than the broader labor market, and that watching them closely provides early warning of shifts to come.
The other two dashboards widen the lens. The Takeoff Tracker monitors a dozen macroeconomic indicators tied to AI advancement, watching for signs that the technology's effects are accelerating across the economy. The Adoption Monitor draws on surveys and international data to gauge how widely AI is actually being used, with some surveys cited by the lab showing self-reported generative AI adoption approaching 50 percent. Together the three create a layered picture, from the leading-edge occupations most exposed, to the broad macroeconomic signals, to the on-the-ground reality of how many people are actually using the tools. No single number could capture a change this multifaceted.
The Questions That Matter
What gives the project intellectual coherence is the set of questions it is organized around. The platform addresses five core questions spanning employment and wage effects, productivity, consumer benefit, evolving skills, and the crucial issue of labor substitution versus complementarity, whether AI replaces workers or makes them more productive. That last distinction is the one that matters most for how societies should respond, because a world where AI substitutes for human labor demands very different policy than one where it complements and amplifies it. Yet it is precisely the question on which we have the least reliable evidence.
By structuring its measurement around these questions, the lab is attempting to move the AI labor debate from speculation toward evidence. Much of the current discourse consists of confident predictions, optimistic or apocalyptic, that are largely untethered from data because the data did not exist. A platform that systematically gathers evidence on substitution versus complementarity, on which skills are gaining or losing value, on where productivity is actually rising, offers a path out of the dueling-anecdotes phase of the conversation. Whether the data ultimately supports the optimists or the pessimists, having the data at all is a substantial improvement over arguing in its absence.
A Public Good in a Contested Debate
There is real value in the fact that this platform is freely accessible and produced by a university rather than sold by a vendor with a stake in the answer. The debate over AI's economic effects is heavily contested and heavily interested, with technology companies inclined toward optimism about productivity and various critics inclined toward alarm about displacement. An independent, transparent, publicly available source of measurement provides a common factual foundation that all sides can reference, which is exactly what a polarized and high-stakes debate needs. The project is supported by Schmidt Sciences, the Siegel Family Endowment, and individual donors, funding that keeps it outside the commercial incentives that might skew the findings.
That independence matters because the conclusions drawn from this data will shape decisions about education, workforce policy, and social safety nets, decisions too important to rest on self-interested analysis. When a company with billions riding on AI adoption tells us the technology is purely complementary to human labor, we are right to be skeptical of the source. When an independent academic instrument measures the same question transparently, the findings carry a different weight. The Stanford project's value lies as much in its credibility and openness as in any particular dashboard, because trustworthy measurement is the scarce resource in this debate.
Why Leaders Should Pay Attention
For education and business leaders, the AI Economic Indicators offer something genuinely useful: an early-warning system for the labor effects of AI that does not require waiting for official statistics to belatedly confirm what is already underway. Educators designing curricula need to know which skills are gaining and losing value before they have committed years to teaching the wrong ones. Employers planning workforce strategy need to see where substitution is actually happening rather than where it is merely feared. A real-time, evidence-based view of these shifts is a planning tool, not just an academic curiosity.
The deeper significance is what the project's very existence signals about the state of our knowledge. We are living through an economic transformation that we lack adequate tools to measure, and that gap between the speed of change and the speed of our understanding is itself a risk. Initiatives like this one are attempts to close that gap, and they deserve attention and support from everyone whose decisions depend on understanding where AI is taking the economy of work. We would encourage leaders to treat the Stanford indicators as a resource to consult regularly rather than a headline to note once, because in a transformation this fast, the organizations that see the shifts early will adapt while others are still waiting for confirmation.
