By Lina Irawan 
Acemoglu, a towering figure in economics, a Nobel laureate, and co-author of influential works like Why Nations Fail, finds himself increasingly exasperated by the polarized and often unproductive discourse surrounding artificial intelligence. While he readily acknowledges the profound dangers posed by unchecked technological advancement – a concern he has voiced for years – his current frustration stems less from the technology itself and more from the inability of society to engage with it in a balanced, pragmatic manner. In a recent interview with Fortune, Acemoglu articulated a nuanced position that defies easy categorization, positioning him as a critical observer of a debate he believes is crippled by extremes.
With a blend of intellectual rigor and weary resignation, Acemoglu critiques the prevailing "camps" in the AI discussion, evoking the cynical wit of Groucho Marx: he wouldn’t want to belong to any club that would have him as a member. This sentiment underscores his rejection of the simplistic narratives that dominate the public discourse, which he argues are hindering meaningful progress and thoughtful policy development.
On one side, he identifies the "true believers"—a group he labels as "quasi-moderate-friendly" types. These individuals are so thoroughly convinced of AI’s inherent benevolence and its capacity to deliver universal good without exception that any challenge to this utopian vision "drives them insane." This camp, often comprising tech evangelists, venture capitalists, and certain researchers, champions AI as the panacea for societal ills, promising unprecedented productivity gains, medical breakthroughs, and a new era of prosperity. They frequently highlight advancements in areas like drug discovery, climate modeling, and personalized education, dismissing concerns about job displacement, ethical dilemmas, or power concentration as mere Luddite anxieties or temporary growing pains. Their unwavering optimism, Acemoglu implies, blinds them to the genuine risks and inequities that AI could exacerbate, leading to a defensive posture against any form of critical inquiry.
Conversely, Acemoglu points to the skeptics who, in his view, refuse to acknowledge AI’s genuine capabilities. This camp often dismisses sophisticated AI models as mere "stochastic parrots"—systems that generate plausible-sounding text or responses based on statistical patterns learned from vast datasets, rather than possessing true understanding or intelligence. Prominent critics, often from linguistic or philosophical backgrounds, argue that such models lack genuine comprehension, agency, or consciousness, and that anthropomorphizing them is a dangerous oversimplification. Acemoglu notes that "There are people on the left – or part of the left – that just will refuse any argument that says AI has capabilities, it just drives them insane." He sighs, lamenting the intellectual gridlock: "it’s very unproductive." This refusal to credit AI with any authentic advancement, he contends, is as unhelpful as uncritical embrace, leading to a denial of the transformative power, both positive and negative, that AI genuinely holds.
Acemoglu firmly rejects both extremes. "I think you have to really have your head in the sand to think that AI is a stochastic parrot right now," he states, arguing that anyone who actively uses and experiments with current models will quickly discern their significant, if imperfect, capabilities. Yet, his acknowledgment of AI’s power is not an endorsement of unbridled optimism. "But I’m also not willing to go along with some inchoate belief that everything will work out fine." This refusal to settle for either extreme—blind faith or absolute dismissal—is central to his call for a more mature and complex engagement with AI.
He readily concedes that "frontier models" are indeed making genuine advances. He points to their impressive strides in comprehension, sophisticated coding, and even groundbreaking scientific and mathematical discoveries, noting, for instance, "it’s really great, the way proofs are being done." This refers to breakthroughs in areas like formal verification using AI-assisted proof assistants, which are revolutionizing fields from computer science to pure mathematics. These tangible achievements, Acemoglu argues, should be celebrated for their potential to unlock new knowledge and drive innovation. However, this celebration must be tempered by a sober concern for potential negative consequences, such as future job displacement, increased inequality, and ethical quandaries. He emphasizes the urgent need for society to simultaneously hold these "two apparently conflicting ideas in your head at the same time." This capacity for complex thought, he laments, "has become sort of radical" in contemporary discourse. This intellectual climate, he suggests, makes the timing of his book, What Happened to Liberal Democracy?, particularly apt.
Acemoglu views the current moment as a time of profound crisis, and the terms of the AI debate merely mirror the deep damage inflicted upon our political system over an extended period. The anger, polarization, and inability to find common ground in AI discussions reflect a broader societal malaise.
The Escalator Effect and the Jagged Frontier
"We live in an environment that’s been partly shaped by social media," Acemoglu explains, "and there is a tendency to escalate everything, because that gets attention. Politics is like that. The other topic like that, unfortunately, is AI." This "escalator effect" transforms complex issues into sensationalized binaries, preventing the nuanced deliberation necessary for effective policymaking.
The economist draws a direct connection between this escalating rhetoric and the public’s inability to simultaneously acknowledge AI’s real capabilities as a likely general-purpose technology and its potentially destructive social and economic effects. This reflects a political culture that has lost its capacity to deliberate over trade-offs, build common ground, and direct economic change toward shared ends – functions that were once hallmarks of a healthy liberal democracy.
He references Ethan Mollick’s concept of the "jagged frontier" of AI capabilities, an apt description for a technology that excels remarkably at some tasks while proving unreliable or even nonsensical at others. AI can be exceptional at generating code or analyzing vast datasets, yet it might struggle with nuanced human interaction or exhibit biases embedded in its training data. This means that, in many practical applications, "You need to do a lot of detailed babysitting," Acemoglu notes, underscoring the gap between AI’s impressive demonstrations and its seamless integration into the workforce. Beyond coding, he flags a lack of widespread adoption, observing that customer service employment has barely budged in recent years, a similar trend seen in manufacturing. This challenges the narrative of imminent, widespread job destruction, suggesting a slower, more uneven integration than often predicted by either extreme camp.
Yet, Acemoglu maintains a cautious hope. In his Groucho-esque fashion, he suggests that this messy, imperfect state of things, this "jagged frontier," might serve a useful purpose. "I wouldn’t call myself an optimist, I would say I resolutely refuse to give up hope." This refusal to despair is rooted in the belief that the current limitations of AI, coupled with a societal willingness to engage thoughtfully, still allow for the possibility of shaping its trajectory towards more beneficial outcomes.
Two Economists Talking: A Model for Debate
Acemoglu points to his long-standing professional relationship with his former colleague and fellow star economist, Stanford’s Erik Brynjolfsson, as a rare model for how the AI debate could proceed. The two have publicly and sharply disagreed about AI’s immediate impact on productivity, with Brynjolfsson often projecting substantially greater and swifter gains than Acemoglu. Brynjolfsson, known for his work on the "productivity paradox" (the idea that new technologies like IT or AI take time to translate into measurable economic growth), often argues that we are on the cusp of a productivity explosion that current statistics simply haven’t captured yet. Acemoglu, while acknowledging AI’s potential, has been more cautious, emphasizing that productivity gains depend heavily on how the technology is developed and deployed.
Despite these intellectual divergences, Acemoglu states, "Erik and I actually agree on many things." Their ability to engage in respectful, evidence-based disagreement, while still identifying common ground and shared goals, is precisely what he wishes to see more of in policy circles. He expressed pleasure that Brynjolfsson has increasingly echoed the call for redirecting AI in more human-complementary and human-friendly ways, a cause Acemoglu says has "been my bugbear for almost two decades." This concept of "human-complementary AI" is crucial, advocating for technological design that augments human capabilities, creates new tasks, and enhances worker value, rather than simply automating existing tasks to displace labor.
Acemoglu also acknowledges Brynjolfsson as "the main scholar showing the potential job losses from AI," citing his work, often in collaboration with Stanford’s John J. Horton and supported by ADP data, which has extensively documented the disproportionate impact of AI on entry-level jobs, particularly for women, through the "Canaries dashboard." Fortune has reported extensively on these findings.
Acemoglu’s own engagement with AI explicitly dates back to at least 2018, but his foundational research on the underlying question—whether new technologies replace workers or create new tasks that raise their value—spans much longer. His seminal work with Pascual Restrepo developed a robust framework for understanding automation as a dual force that can simultaneously displace labor and generate new work. This ambiguity lies at the heart of today’s AI debate, informing discussions around the "lump of labor fallacy" (the mistaken belief that there’s a fixed amount of work to be done, so new technology automatically means less work for humans) and the "Jevons paradox" (where increased efficiency in resource use, like labor through automation, can lead to increased overall demand and thus potentially more rather than less work or resource consumption). His 2024 Nobel Memorial Prize in Economic Sciences, shared with Simon Johnson and James Robinson, concerned their pioneering research on how institutions and power structures shape prosperity, or fail to, providing a critical lens through which to view the governance and direction of AI.
The point, Acemoglu stresses, is that there are no easy answers. Neither he nor Brynjolfsson is a straightforward technological booster or a Luddite skeptic. Instead, they represent a commitment to moving toward a clearer understanding of the complex trade-offs, potential problems, and viable solutions.
Liberalism’s Broken Bargain and Political Dysfunction
According to Acemoglu’s book, What Happened to Liberal Democracy? (co-authored with Simon Johnson), liberal democracy historically relied on more than just elections and constitutional rights. "Shared prosperity" was the essential "glue," the "main promise" that bound the system together. However, somewhere in the transition to what he terms the "postindustrial economy"—a shift marked by deindustrialization, the rise of the service sector, skill-biased technological change, and increasing globalization—that fundamental bargain disintegrated.
The resulting societal divides, Acemoglu argues, are precisely those observed in the dysfunctional AI debate. The educated and less educated became increasingly separated, with the more educated gravitating towards the center-left, while the working class, often left behind by economic shifts, increasingly aligned with the center-right or even the hard right. This stratification has led to a "significant divergence in values" between these camps, coupled with a pronounced gap in "connections and empathy." This chasm fosters what he describes as both "sins of omission" (the center-left’s silence or inaction as inequality grew) and "sins of commission" (cultural politics that actively divide).
He contends that the center-left’s failure to address rising economic inequality, allowing it to grow unchecked, has created a deep class divide. Concurrently, a focus on cultural politics has further alienated segments of the working class, severing the communal roots essential for liberal democracy’s stability. This breakdown, Acemoglu asserts, has fostered an environment where complex, shared challenges like AI are impossible to address constructively.
The DSA and the Perils of Cultural Politics
Acemoglu’s framework also informs his ambivalent view of groups like the Democratic Socialists of America (DSA) and figures like Zohran Mamdani, the telegenic New York City legislator known for elevating local politics to national prominence. The DSA, Acemoglu believes, is a "mixed bag." He credits them for successfully highlighting the crucial theme of affordability and economic justice, which resonates with many. However, he criticizes their approach as "preaching to their base," noting their limited impact with Black voters, who have often responded lukewarmly to their messaging. Furthermore, he argues they have "doubled down on cultural politics… exactly the kind of policy ideas and rhetoric that alienate the working class [and] creates an adversarial attitude."
He clarifies that he isn’t necessarily disagreeing with the DSA’s stances on certain issues, but rather with their methods. He advocates for leveraging liberal democracy’s mechanisms for consensus-building, drawing examples from his book, such as state-level referenda on gay marriage or grassroots movements for abortion rights in Ireland. These bottom-up processes allowed for public deliberation and persuasion, fostering genuine shifts in opinion rather than imposing top-down solutions. "In both cases the evidence is that many people change their views," he said, emphasizing that the process cannot be forced. Liberalism, he argues, should not be a "cookbook" dictating answers to every controversial question but rather a framework for resolving difficult issues through persuasion, broad public participation, and compromise—a challenging feat in a time of crisis.
Regarding democratic socialism itself, Acemoglu finds it "a very amorphous concept" currently. He is not inherently opposed to specific Mamdani policies, such as a wealth tax or a pied-à-terre tax. However, he questions the efficacy of implementing such measures at a city level rather than nationally, explaining that state-by-state wealth taxes can trigger competition among jurisdictions, leading to capital flight and benefiting states like Texas and Florida that offer more favorable tax environments.
Midterm Elections and the Future of AI
When the conversation shifts to Donald Trump and the upcoming midterms, Acemoglu maintains his characteristic, not-quite-optimistic outlook. "I still think we’re going to have elections in November and both sides will get counted," he observes, a cautious statement in an era of heightened electoral mistrust. However, looking further ahead, he admits, "Two years from now, I have no idea," reflecting profound uncertainty about the durability of democratic norms.
AI anxiety, particularly concerns about job displacement and the rapid development of data centers (which consume vast amounts of energy and land), has emerged as a significant issue in the current election season. Acemoglu’s book calls for a paradigm shift towards "pro-worker AI" – a technological development path that prioritizes augmenting human capabilities and creating new tasks over simple automation and displacement. This vision is coupled with a demand for a stronger safety net and greater redistribution of wealth.
To achieve "pro-worker AI," Acemoglu suggests concrete policy changes: supporting AI tools that genuinely increase the effectiveness and wages of a broad spectrum of workers; reconsidering tax rules that often favor capital gains over wages, thus incentivizing automation over human labor; limiting the dominance of big tech companies to foster competition and prevent monopolies; and, crucially, giving workers a stronger voice at the table through mechanisms like collective bargaining or co-determination.
Above all, he stresses, the unthinkable must be avoided: "If 50, 60, 70% of the people become jobless, hopeless, feeling dispensable, having no dignity at work, then I don’t think we can have a liberal democracy society." The economic disenfranchisement of a large portion of the population, he warns, is an existential threat to democratic institutions.
Asked about the implications of Trump potentially steering AI policy until the next presidential election, Acemoglu concedes, "It’s not going to change radically for two years," but quickly adds a critique of the other side, "I wouldn’t say Democrats are on the ball, either." His prescription, therefore, remains consistent: more of his beloved liberal democracy—not as a fixed ideology, but as a dynamic process of negotiation, persuasion, and compromise.
He reflects on the remarkable achievement of building advanced societies, capable of negotiating conflicts and reaching compromise without routine descent into violence. "We come from very cantankerous apes," Acemoglu muses, marveling at "amazing" human constructs built from this "raw genetic inheritance." The thought of losing liberal democracy, he concludes, would be "so devastated."
"We need to redirect AI," he reiterates, expressing deep concern that "things won’t work out if we don’t change things now." The imperative, in these crisis-ridden times, is to truly listen to one another, to cultivate the intellectual fortitude to hold two conflicting ideas simultaneously, and to overcome our inherent "cantankerous natures." As Acemoglu himself acknowledges, this is undoubtedly easier said than done.
