Artificial intelligence promises massive productivity gains, but not everyone can immediately jump in to reap its benefits. While some firms and countries race ahead with new tools, others deliberately stick with older technologies, a choice may be entirely rational.

In this episode of the Fuqua Insights podcast, Professor John Coleman of Duke University’s Fuqua School of Business joins host Sarah Kern to discuss how economies and businesses choose technologies based on the skills of their workforce. Drawing on his paper The World Technology Frontier, co-authored with Francesco Caselli of the London School of Economics, Coleman explores why countries with access to the same global technologies nonetheless adopt very different production methods and experience sharply different income levels.

Coleman’s central finding is that technology adoption is often an optimal choice, not a failure. Coleman distinguishes between skilled and unskilled labor, economic terms based primarily on formal education levels rather than the actual difficulty or value of work. In this framework, "skilled" workers have training that allows them to adapt to and leverage new technologies, while "unskilled" refers to workers with less formal education who may be highly capable but less equipped to use tools like AI.

Advanced economies tend to develop and use technologies that favor skilled labor, while less-developed economies rely on older, less skill-intensive tools because those technologies better match their labor force. As Coleman explains, a poorer country may “optimally choose to adopt what seems to an advanced economy to be a backward technology”, because using more advanced tools without the right skills can actually reduce productivity.

While technologies developed a long time ago are most suitable for labor that is mostly unskilled, newer technologies are designed for highly educated workers, Coleman notes. This framework applies directly to AI: “AI was developed by advanced economies that have the skilled labor that would benefit most from AI,” he says. Without that skill base, adopting AI may not pay off.

Business leaders should remember to align innovations with workforce capabilities. “Adopting AI without a well-developed plan to integrate it into your workforce would likely be a disaster,” Coleman warns. Globally, AI may widen gaps in the short run, but Coleman emphasizes that technology is not a zero-sum game. Over time, moving up the technology frontier can create opportunities for growth across economies.

Duke Fuqua Insights features digestible conversations with our faculty about the most impactful research from their careers, including studies they teach in Fuqua classes. New episodes every other week in season.

For more from Duke Fuqua, visit us on LinkedIn, Instagram, Facebook, Bluesky, and the Duke Fuqua Insights newsletter.

Professor David Brown explains how simple strategies can guide better decisions even when information is incomplete

New job postings appear daily. Real estate markets update constantly with fresh listings. In an environment where alternatives continuously multiply and options can seem endless, the hardest decision is knowing when to stop searching and commit.

In this episode, David Brown, the Snow Family Business Professor of Decision Sciences at Duke University’s Fuqua School of Business, discusses how people and organizations can make better decisions when information is scarce or costly.

Building on economist Martin Weitzman’s classic “Pandora’s Box Problem,” Brown and his co-author, Fuqua Ph.D. student Cagin Uru, found that straightforward search rules perform nearly as well as complex algorithms. Their research shows a surprisingly simple solution: commit upfront to search a specific number of alternatives based on search costs, then simply rank what you've seen and choose the best.

What makes their approach practical and appealing is its simplicity: it requires only the ability to rank alternatives you've seen and the discipline to stop searching at the right point, not probability calculations or complex data analysis. This applies broadly, from navigating job searches to booking flights to hiring contractors.

The conversation also explores when sophisticated algorithms are truly necessary. Their research shows that, across several search settings, their simple, transparent rules perform nearly as well as those based on more complex approaches (e.g., AI), raising questions about when algorithmic solutions are worth the investment.

Duke Fuqua Insights features digestible conversations with our faculty about the most impactful research from their careers, including studies they teach in Fuqua classes. New episodes every other week in season.

For more from Duke Fuqua, visit us on LinkedIn, Instagram, Facebook, Bluesky, and the Duke Fuqua Insights newsletter.

A single decision improved by 1% might seem trivial. But make 300 small improvements over a year, and the compounding effect becomes transformative. A/B testing allows companies to systematically test different approaches and optimize performance, but research shows that the startups that could benefit most are the least likely to use it.

In this episode, Professor Sharique Hasan of Duke University’s Fuqua School of Business discusses his paper “Experimentation and Start-up Performance: Evidence from A/B Testing,” which focuses on how startups use A/B testing to drive performance. Based on data from more than 35,000 startups, Hasan and his coauthors found those that adopt A/B testing experience significantly higher performance over time—sometimes doubling outcomes after a year.

Hasan explains that while the impact is strongest for smaller and non–Silicon Valley startups, these firms often lack the resources to implement A/B testing effectively. For them, he introduces the concept of “experimental thinking” as a more accessible alternative: a mindset of comparing options rigorously, asking the right causal questions, and framing decisions with clear counterfactuals.

Drawing from both large-scale quantitative analysis and rich qualitative insights from tech practitioners, Hasan describes how small, compound decisions can lead to transformative outcomes.

Duke Fuqua Insights features digestible conversations with our faculty about the most impactful research from their careers, including studies they teach in Fuqua classes. New episodes every other week in season.

For more from Duke Fuqua, visit us on LinkedIn, Instagram, Facebook, Bluesky, and the Duke Fuqua Insights newsletter.

Professor Bill Mayew explores whether public companies have visibility into the macroeconomy to filter errors in GDP data—and what that means for economic forecasting

Every quarter, the U.S. Bureau of Economic Analysis releases its initial GDP estimate—a flagship measure of economic health that influences corporate boardrooms, Federal Reserve policy, and investor portfolios. But there’s a catch: these early numbers are often wrong.

In this episode, Professor Bill Mayew of Duke University’s Fuqua School of Business discusses his research, published in the Journal of Accounting and Economics, on how corporations respond when government economic data contains errors. Mayew explains why concerns about a potential “macro data crisis” have gained traction and why errors in economic data are not necessarily signs of dysfunction.

Initial GDP estimates rely on incomplete survey data—less than half from actual three-month surveys—with the rest from extrapolations. The Bureau of Economic Analysis refines these estimates at the one-year and five-year marks as more data arrives. Revisions are therefore expected and necessary.

Mayew’s research examined whether large public companies with a unique pulse on the economy could see through the errors inherent in initial GDP estimates. Analyzing firm-level behavior, he and his coauthors found firms tend to take preliminary GDP figures at face value, failing to filter out the inherent noise. When GDP data signals strength in one quarter, companies increase investment, production, and inventory the next — and the same pattern occurs whether the GDP signal reflects real economic change or statistical error.

For policymakers, the findings underscore the need for caution when substituting government data with private sector sources like ADP payroll information. While private data may complement government releases in some cases, Mayew emphasizes government data from agencies like the Bureau of Economic Analysis and Bureau of Labor Statistics still has substantial value.

Instead, he concludes, “we need to think of other ways to improve government data, which may be increasingly possible as new and creative ways of measuring economic activity occur.”

Duke Fuqua Insights features digestible conversations with our faculty about the most impactful research from their careers, including studies they teach in Fuqua classes. New episodes every other week in season.

For more from Duke Fuqua, visit us on LinkedIn, Instagram, Facebook, Bluesky, and the Duke Fuqua Insights newsletter.

Professor Tong Guo explains how mandated transparency didn't reduce pharmaceutical payments to physicians—instead, it taught companies to optimize them.

When the federal government mandated that pharmaceutical companies publicly disclose every payment to physicians—from conference sponsorships to consulting fees—policymakers expected transparency to reduce potential conflicts of interest. Instead, the payments kept flowing, and companies learned to optimize them.

In this episode, Professor Tong Guo, an associate professor of Marketing at Duke University’s Fuqua School of Business, discusses her study of the Sunshine Act—a federal law requiring pharmaceutical and medical device companies to publicly disclose payments to healthcare providers. Published in the Journal of Marketing Research (2021), Guo's research using advanced machine learning methods called causal forests analyzed $100 million in payments between 16 antidiabetic brands and 50,000 physicians. Her findings reveal a nuanced reality: while total payments did not decline significantly, they shifted toward physicians who prescribe more expensive drugs and generate higher ROI for firms.

As Guo explains, most disclosed payments are legal, including sponsorships for events, conference travel, and educational presentations. "Much of these expenditures are considered legal," she notes, "so it's natural that it doesn't come with much pressure to cut it down."

"For firms, the number one rule for them to run their business is always to think about their ROI model," she explains. The transparency regulation gave firms information about which competitors were reaching out to which physicians and when, allowing them to optimize their existing relationships for maximum return. When transparency gives all competitors access to the same information, firms don't retreat—they optimize.

For MBA students and professionals, Guo's findings offer critical lessons extending beyond healthcare. Transparency doesn't always lead to restraint. Understanding who benefits from newly available information—and how—is essential across industries, from healthcare to digital marketing. As Guo points out, similar disclosure regulations now apply across industries —from TikTok influencers required to disclose brand sponsorships to financial services and beyond. "Transparency regulations would not necessarily lead to drastic changes of how people practice their business," she says. Different parties have different capabilities to leverage disclosed information, potentially creating new competitive advantages rather than leveling the playing field.

Duke Fuqua Insights features digestible conversations with our faculty about the most impactful research from their careers, including studies they teach in Fuqua classes. New episodes every other week in season.

For more from Duke Fuqua, visit us on LinkedIn, Instagram, Facebook, Bluesky, and the Duke Fuqua Insights newsletter.

Professor Ali Makhdoumi reveals why your friend's social media activity might be compromising your privacy, even when you share nothing at all

Every time your colleague shares their location data or a friend posts their workout routine, they're inadvertently exposing details about you–even if you've never agreed to share your data. This hidden web of data spillovers means companies can predict your preferences, behaviors, and personal information simply by analyzing the digital footprints of people in your network.

In this episode, Professor Ali Makhdoumi of Duke University's Fuqua School of Business discusses his research on personal data markets, based on his paper "Too Much Data: Prices and Inefficiencies in Data Markets," co-authored with 2024 Nobel Prize winner Daron Acemoglu. He explains that what we think of as personal, private data is actually more like a public good. Platforms can infer your information indirectly through your connections, creating what economists call "data externalities."

Makhdoumi explores why current data markets are so structurally inefficient. When your data can be predicted from others' sharing decisions, you lose bargaining power and companies acquire personal information at depressed prices. This creates market dynamics where users share more data than is socially optimal, often receiving compensation that doesn't reflect the full social costs.

The implications extend beyond individual privacy concerns. Makhdoumi's research shows that under certain conditions, shutting down data markets entirely would improve societal welfare. For business leaders, this challenges conventional thinking about data as a valuable corporate asset and raises questions about sustainable data strategies.

Makhdoumi proposes innovative solutions, including "decorrelation" techniques that could allow beneficial data sharing while protecting privacy. He also outlines policy approaches that could help realign market incentives with social benefits. The research offers a framework for companies thinking more strategically about data acquisition, user trust, and the long-term sustainability of data-driven business models.

Duke Fuqua Insights features digestible conversations with our faculty about the most impactful research from their careers, including studies they teach in Fuqua classes. New episodes every other week in season.

For more from Duke Fuqua, visit us on LinkedIn, Instagram, Facebook, Bluesky, and the Duke Fuqua Insights newsletter.

Ronnie Chatterji, Professor at Duke University's Fuqua School of
Business and Chief Economist at OpenAI, joins Jenny Laurence, MBA '26 to discuss his recent research paper analyzing over a million ChatGPT conversations to uncover patterns in where artificial intelligence is making impacts in our homes, our work, and our lives.

Duke Fuqua Insights features digestible conversations with our faculty about the most impactful research from their careers, including studies they teach in Fuqua classes. New episodes every other week in season.

For more from Duke Fuqua, visit us on LinkedIn, Instagram, Facebook, Bluesky, and the Duke Fuqua Insights newsletter.

Trump's push to end quarterly reporting could reshape American business. Professor Rahul Vashishtha explains what research shows about the trade-offs.

When companies report earnings more frequently, they make different investment choices, often abandoning profitable long-term projects that don't pay off quickly. This behavioral shift sits at the heart of President Trump's renewed call to end quarterly reporting requirements in favor of six-month reporting cycles.

In this episode, Professor Rahul Vashishtha discusses his research examining what happened during the historical shift from annual to semi-annual to quarterly reporting between 1950 and 1970.

Vashishtha found that when companies were required to report more often, they significantly reduced their investments in long-term projects. More concerning, this investment decline was accompanied by lower productivity, reduced sales growth, and weaker financial performance. This suggests companies weren't just eliminating waste, but abandoning profitable opportunities.

This "managerial myopia" was most pronounced in industries where investments take years to pay off, precisely where quarterly earnings reports are least effective at capturing true value creation. As Vashishtha explains, "When you start increasing the frequency of your performance measures, what you do really is create a premature evaluation of decisions which are best considered over a much longer horizon."

The episode explores both sides of the reporting frequency debate, examining the trade-offs between transparency and long-term value creation. Vashishtha also offers practical advice for corporate leaders and investors on encouraging long-term thinking, including cultivating patient capital, strategic communications, and thoughtful incentive design.

Record date: September 16, 2025

Duke Fuqua Insights features digestible conversations with our faculty about the most impactful research from their careers, including studies they teach in Fuqua classes. New episodes every other week in season.

For more from Duke Fuqua, visit us on LinkedIn, Instagram, Facebook, Bluesky, and the Duke Fuqua Insights newsletter.