Nearly 30 years ago, as the dean of what was then known as the Tucker Foundation at Dartmouth, I wrote a series of articles on the growing economic and class divide in our country. I worried then that the trajectory of income and wealth disparity, if unchecked, would “add strain to our lives and the lives of our children and grandchildren.”
A generation has passed, and, while national wealth has increased, the economic divide has also widened. In 1978, the year I graduated from Dartmouth, the top 0.1 percent of Americans owned 7 percent of the nation’s wealth. By 1996, when I was at the Tucker Foundation, that share had doubled to 14 percent. Today, the super-rich control more than 20 percent of our wealth.
Meanwhile, the share owned by the bottom 50 percent of the population has fallen from 3.5 percent in 1996 to 2.5 percent in 2026. Indeed, the bottom 8 percent of the U.S. population has a negative net worth, with their debt exceeding their marketable wealth.
Today, AI threatens to widen the gap still further. Dario Amodei, the CEO of leading AI company Anthropic, painted a picture of a bifurcating economy during a speech at the World Economic Forum in January at Davos, Switzerland. He warned that the technology may lead to a top economic tier inhabited by uber-wealthy manipulators of AI, primarily based in Silicon Valley, and, for the rest of us, a second tier that sinks into economic stagnation and unemployment.
That same day in Davos, BlackRock CEO Larry Fink, arguably the most powerful asset manager in the world and chair of the forum, noted that the wealth associated with AI is “flowing to the owners of models, data, and infrastructure,” while the rest of us are “spectators watching it happen.” Indeed, Fink noted, the most significant power of AI may be its power to concentrate wealth to a degree more “than any healthy society can sustain.”
AI will deepen geographic inequality, too. The centers of technology and AI adoption are concentrated on the East and West coasts, states that already have the highest median income. All 20 states with the lowest incomes are in the South or Midwest, where many people make a fraction of what coastal residents earn. Meanwhile, the median compensation for individual employees at San Francisco-based OpenAI in 2026 exceeds $1.3 million, including the value of stock options.
As AI companies go public, the scale of wealth concentration is becoming even more evident. The SpaceX IPO in June minted thousands of millionaires among the ranks of employees, with 400 earning more than $100 million. Hundreds of venture capitalists and investors will walk away with millions more. The market expects similar riches from the IPOs of Anthropic, OpenAI, and dozens of AI infrastructure companies.
The global rupture may be even more dramatic. AI adoption is accelerating in the United States, China, and Europe but lagging badly in the global south. Research at the International Monetary Fund in 2025 concluded that “AI will exacerbate cross-country income inequality, disproportionately benefiting advanced economies.” To the extent that migration is driven by poor people seeking better economic opportunities, today’s problems may pale in comparison to the immigration pressure we face in future years.
Throughout history, concentrated wealth has distorted political power. An elite aristocracy controlled the reins of power in Europe for centuries before two world wars decimated its sources of wealth and scrambled social structures. Most of the countries in Central America were controlled by a small number of land-owning families for decades until the revolutions of the post-WW II era.
The United States has relied on its separation of powers and the strength of political institutions to check the power of wealth. At the dawn of the 20th century, Teddy Roosevelt took on the trusts. In the 1930s, FDR reined in the banking and securities industries and rewrote the social contract to protect the poor and the aging. In the 1960s, Lyndon Johnson sought not just to manage inequality but to also eliminate poverty. In each case, popular democratic opposition, judicial action, and regulatory reform tempered the power of concentrated wealth.
Today, our governing institutions have shown neither the ability nor the inclination to regulate AI in a manner that might mitigate the divisiveness and harmful potential of the technology. The pace of AI change outstrips the capacity of our legislative and judicial institutions to understand, let alone regulate, the industry. Government agencies can’t afford to hire those who might have the necessary expertise, and early executive actions have shown a preference for limited regulation to ensure dominance over Chinese competition. Meanwhile, tax policy has increasingly favored the wealthy for decades, especially those reaping gains on capital.
Without regulation, AI may concentrate not only wealth but also the levers of democratic government through its capacity to shape what we see, how we think, and what we believe. The combined power of wealth and influence in the same few hands seems certain to have a corrosive impact on our electoral system.
The framers of our Constitution understood well the danger of concentrated wealth and influence. John Adams argued that the institutions of government must control the “ambition and avarice” of the wealthy, warning that if their power remains unchecked “they will destroy all equality and liberty.” He argued further that the wealthy, unless controlled, “will strengthen themselves by insensible degrees, by playing into each other’s hands more wealth and popularity, until they become able to govern elections as they please and rule the people at discretion.”
What’s different now is the ability of AI to concentrate that wealth and power to an unprecedented degree. This is not just an economic problem; it’s also a political and moral problem. Today, many would argue our political foundations are crumbling and our moral compass can no longer find a common true north. The corrupting influence of wealth and power in the hands of the few is the most enduring lesson of political history. It is not a coincidence that the majority of Americans believe our country is headed in the wrong direction, while at the same time the stock market is reaching record levels and the number of billionaires grows exponentially.
What does this have to do with Dartmouth? Thirty years ago, I argued that Dartmouth should encourage thought and discussion about the growing wealth gap in our country and the social and political consequences of such a divide. I encouraged College leaders and students to include in Dartmouth’s mission a focus on moral and ethical education, not just academic excellence. As John Sloan Dickey, Dartmouth’s 12th president, wrote in 1951 when he founded the Tucker Foundation, “To create the power of competence without creating a corresponding sense of moral direction to guide the use of that power is bad education.”
Today, the power of AI brings the matter into higher relief. Some creators of AI, like the creators of nuclear weapons, are asking themselves whether they have unleashed a power that will run amok without moral direction and control. It is here that Dartmouth has a special role to play. Seventy years ago, during the summer of 1956, Dartmouth hosted a two-month workshop for computer scientists who coined the term “artificial intelligence” and launched the field that now shakes the foundations of our economy, society, and politics.
Now, Dartmouth is positioning itself to be a leader in the AI revolution, championing its partnership with Anthropic, establishing an AI-intensive curriculum at the Thayer School of Engineering, and developing AI-focused courses in disciplines as diverse as film studies and economics. Teaching students to apply AI in their fields of study and research is important and giving them the tools to compete for jobs after graduation is important, too.
However, competence without moral direction, to paraphrase Dickey, is not enough. Students need to learn how to wield the enormous power of AI responsibly—what experts call AI literacy. That means more than implementing standards for cheating and essay writing. It means making the moral and ethical use of AI a core part of a Dartmouth education.
In the 1950s, Dartmouth nurtured the ideas that laid the groundwork for AI. It can lead again by turning out graduates who are prepared to control AI so it doesn’t control us.
Scott Brown was the dean of the Tucker Foundation at Dartmouth from 1996 to 1999. He lives in Hanover.
The views expressed in this opinion article are the author's own.
