Change Is Certain. Progress Is Not. Why We’re Building a People-Centered AI Agenda.

By Evan LeFlore, Director of AI, Innovation, and Economic Opportunity, Better Markets

An introduction to our new work at Better Markets and why the public interest needs a seat at the table while the rules for AI are still being written.

Soon, an AI system you’ll never see and can’t question will decide whether you get a loan, what your interest rate is, whether your insurance claim is paid, and whether your job still exists. That’s not a forecast about 2035. It’s happening now, in finance, retail, logistics, and the office jobs that have long been a path into the middle class. The decisions are being made faster, at greater scale, and with less transparency than anything that came before, and the people most affected by them have the least say in how they’re made.

That gap is the reason for this work. At Better Markets, we have a phrase that captures the whole problem: change is certain, progress is not. The two are not the same thing and treating them as if they were is how working families end up paying for transformations they never agreed to. Whether AI makes life better or worse for most people depends entirely on the rules written around it, and right now those rules are being drafted by the firms racing to deploy AI first and worry about the consequences later. The key is to encourage and support change while harnessing and channeling it so that change becomes progress that is real, broadly shared, and beneficial for the maximum number of people.

We’re writing to introduce a new effort to change that. Better Markets has spent fifteen years as one of the few independent voices in Washington fighting for an economy and financial system that work for Main Street rather than just Wall Street. We’ve now brought that fight to artificial intelligence, with a dedicated program led by Better Markets’ inaugural Director of AI, Innovation, and Economic Opportunity. The aim is straightforward: as AI reshapes American life, keep ordinary people at the center of the decisions around its development, not an afterthought.

The Stakes

Start with the speed. OpenAI’s ChatGPT reached 100 million users two months after launch and claimed over 900 million weekly active users two years later. Nearly every major employer in America, including every large financial institution, is racing to deploy some form of AI, often without the people on the receiving end even knowing a machine is making the call.

That speed is the heart of the problem, because the actors setting the pace have the least incentive to slow down and ask who gets hurt. The firms chasing first-mover advantage and short-term profits are driving AI’s development, and they will readily admit that society’s structures cannot keep up. In that environment, the public interest gets treated as an obstacle to innovation, and left to those incentives alone, the odds of change becoming progress are not high.

This is not a story about good guys and bad guys. Many people inside the industry want AI built to serve people, and they are welcome allies. The point is structural: when the loudest, best-funded voices in the room have the most to gain, someone has to make the case for everyone else. That is the case we intend to make.

The warnings about job losses make clear that this is not abstract, and they’re coming from inside the house. For example, Microsoft’s AI chief Mustafa Suleyman said in an interview with the Financial Times this past February that for white-collar work done at a computer—law, accounting, marketing, project management—“most of those tasks will be fully automated by an AI within the next 12 to 18 months.” Anthropic CEO Dario Amodei has predicted AI could push unemployment to 10–20% within five years. And even JPMorgan’s longtime leader Jamie Dimon has warned the impact of AI-driven job losses could be a “big problem for society” if government and the private sector don’t work together to mitigate the harm.

Even the skeptical framing reinforces the point. Although OpenAI’s Sam Altman conceded that some companies are “AI washing”—blaming AI for layoffs they would have made anyway—he insisted that the “real impact of AI doing jobs in the next few years will begin to be palpable.”

These are the people building and deploying the technology at scale, telling us to prepare.

A People-Centered AI Agenda

The foundation for the policies that will govern AI for years is being laid right now. The voices focused on unfettered development and profit maximization have an army of lobbyists, PR specialists, and well-paid allies with a seat at the table. The public does not. Better Markets intends to fill that empty seat to fight for a people-centered AI economy that works for everyone.

Our agenda is organized around four pillars, each asking the same question in a different domain: does this version of AI strengthen the lives and livelihoods of all Americans, or not? If not, something needs to change.

Pillar 1—Growth & Jobs: Who benefits when AI reshapes the economy? To be clear, AI is generating real value for the American economy. For instance, the Penn Wharton Budget Model estimates that U.S. productivity and GDP levels will be 1.5% higher over the next decade due to AI.

But the gains may not be shared. AI-driven layoffs are already hitting finance, tech, logistics, consulting, media, retail, and manufacturing. The Wall Street Journal has reported that firms like Dow and PayPal will shrink their workforces by roughly 12% to 20%, respectively, even as they pour tens of billions into AI. And the damage also outlasts the layoff: Goldman Sachs research found displaced workers see real-earnings losses of more than 3% upon reemployment and, over the following decade, see earnings grow nearly 10 percentage points less than peers who were never displaced.

To prevent our labor market from splitting in two, this is what we plan to do: track which workers and communities are bearing the brunt and press on whether AI-driven productivity gains actually reach workers and families rather than only shareholders and executives. Shedding light on these questions is the test of whether we build an economy that expands America’s middle class or accelerates its disappearance.

Pillar 2—Credit & Capital: Who gets the money, and on whose terms? Done right, AI in lending, when coupled with other innovations, could open doors. For instance, cash-flow underwriting could expand access to much-needed cash for tens of millions of Americans who are credit-invisible but creditworthy. Done wrong, it entrenches exclusion at greater scale and with less transparency.

The evidence so far is mixed: UC Berkeley researchers found fintech lenders discriminated roughly 40% less than face-to-face lenders on mortgage rates, yet in July 2025 the Massachusetts Attorney General settled with a student lender for $2.5 million after its AI model produced discriminatory outcomes for Black and Hispanic applicants. And the gap is already measurable: the Federal Reserve’s 2024 household economic wellbeing survey found 51% of Black and 44% of Hispanic applicants were approved for less credit than they sought, against 26% of white applicants.

The question is whether AI narrows that gap or automates it. Through this pillar, we will investigate whether AI expands access to affordable, non-predatory credit for the families and small businesses that need it most or entrenches the same patterns of exclusion. This work will also evaluate whether concentrating AI in the largest institutions leaves community banks and the local economies they serve behind.

Pillar 3—Transparency & Accountability: When the machine decides, who is responsible? “Agentic AI” refers to systems that don’t just analyze data but act on it—autonomously making decisions and taking actions, such as approving or denying a loan, with little or no real-time human oversight. Roughly 47% of financial institutions are already testing these agents.

When one denies a claim, rejects an application, or miscalculates risk, our legal system should hold the people who deployed it accountable. Yet when the Fed, OCC, and FDIC issued updated model risk guidance in April 2026, they explicitly excluded agentic and generative AI from its scope, calling these technologies “novel and rapidly evolving.” Banks are now left to self-govern at precisely the moment these systems are making consequential decisions about people’s lives. Closing that accountability gap before the damage compounds is essential, and the alternative is the pendulum-policymaking that swings from laissez-faire to overreaction and serves no one.

Given the importance of transparency and accountability in the rise of agentic AI, we will press for clear answers on which law applies and which agency enforces it when an agentic system gets a consequential decision wrong, whether the consumer-protection and fiduciary standards that govern humans apply to the machines replacing them, and how to build transparency into these systems from the start.

Pillar 4—Transition Gains & Costs: Who pays the bill? Every technological transformation imposes costs, and those costs rarely fall on the people capturing the gains (think of the factory owner who dumped waste in the river while everyone downstream paid the price). The same dynamic is playing out with AI’s buildout.

Data centers demand staggering amounts of electricity, and as Bloomberg reported in its analysis of how AI data centers are sending power bills soaring, areas near them saw wholesale prices surge as much as 267% over five years. A Consumer Reports survey found 78% of Americans worried that data centers will raise their energy bills. Meanwhile, the safety net for displaced workers was never designed for disruption at this speed.

So this is what we plan to do: create visibility on the hidden costs around the infrastructure behind the AI boom, the obligations tech companies owe the communities hosting their data centers, and how a fair share of AI’s gains can be redirected to the ratepayers, workers, and taxpayers now absorbing its costs.

Why We’re Taking a Seat at the Table

Better Markets already understands the economy and the financial system, from its products, culture, and delivery. What we’ve now added is fluency in how AI is actually built and deployed coupled with a working knowledge of which policy levers in Washington produce progress. The result is an advocate that can follow a problem from the engineering, through the financial system, into the regulatory machinery, and back out to the kitchen table—the combination a People-Centered AI Agenda requires.

Where We’re Going

From factory automation in the 1980s to the trade deals of the 1990s to the rise of the internet, history teaches the same lesson every time: the gains went to the top, the costs fell on working families, and government’s response came too late and did too little. This time the disruption reaches beyond the working class, as the jobs of writers, coders, analysts, paralegals, designers, accountants, and even lawyers and bankers are now in the line of fire. As the Wall Street Journal’s Greg Ip has noted, Big Tech has never before caused a job apocalypse, but the open question is whether this is the moment that changes.

Across all four pillars of this AI agenda, we’ll be doing the research, working with policymakers and the public, and demanding the answers to make the case that the public interest belongs in the room while the rules are still being written.

The window to make AI work for Main Street Americans is open. The time to act is now.