AI, Jobs, and Uncertainty: A Leading Expert Weighs In on What We Can Expect

For more than a decade, Thomas Davenport has been one of the leading voices arguing that artificial intelligence would transform work by augmenting, but not replacing, humans.

Today, that outlook is shifting, if still hazy.

“If you’re not worried about AI and the labor market, you’re not paying attention,” said Davenport, the President’s Distinguished Professor of Information Technology and a prolific thought leader on AI in business and the economy. In addition to his role as faculty director of the C. Dean Metropoulos Institute for Technology and Entrepreneurship and co-founder of the International Institute for Analytics, Davenport has written extensively on AI, including The AI Advantage: How to Put the Artificial Intelligence Revolution to Work and All In on AI: How Smart Companies Win Big with Artificial Intelligence.

In a recent interview, Davenport outlined a more cautious view of how AI is reshaping jobs, organizations, and the broader economy.

From Analytical to Agentic

Davenport’s earlier research emphasized collaboration between humans and AI, highlighting real-world examples of employees successfully working alongside AI systems.

“I’ve been relatively optimistic about AI and jobs for over a decade now. I co-authored a book called Only Humans Need Apply that argued that augmentation of people by AI and vice versa was by far the most likely and the smartest way to go,” Davenport said. “That was right for about a decade.”

---

BOOSTING AI SKILLS: Babson teams up with Founderz on AI in Action certificate program.

---

But the emergence of generative AI tools capable of producing text, code, and other content, and newer “agentic” systems that can take action autonomously, has introduced a different level of disruption.

“Before generative AI, AI was mostly about how to make better decisions. How to predict what price to charge, or what customers to make an offer to. I called it analytical AI,” Davenport said. “(Analytical AI) didn’t really replace many jobs, but generative AI, and now agentic AI, are about creating content and performing tasks.”

That shift, he noted, expands AI’s reach into new roles, including knowledge work across industries, from marketing to journalism to software development.

AI and the Future of Work

One of Davenport’s concerns is the future of entry-level workers.

Early-career roles often involve repetitive or structured tasks, the same types of work AI can increasingly handle. That creates a potential bottleneck for talent pipelines.

“If companies don’t hire entry-level workers today, how do you get experienced workers tomorrow?” Davenport said. “We still haven’t figured that out.”

Software development, customer service, and knowledge creation positions also will likely be impacted by AI. However, Davenport cautioned, AI-related job loss estimates are notoriously faulty.

“There have been a lot of predictions of how many jobs would be lost or gained through AI, and the one thing all these predictions have in common is that they’re horribly wrong, and mostly too pessimistic about job loss,” Davenport said. “So far, not much has happened. Will it happen in the future? I think some degree of job loss is likely. “

The Reality Behind AI-Driven Layoffs

Despite recent high-profile announcements linking layoffs to AI, such as Block CEO Jack Dorsey’s dismissal of more than 4,000 employees in mid-March, Davenport cautioned that much of the current job impact is still speculative.

In fact, research he undertook shows that only 2 percent of companies announcing AI-related job cuts actually have the technology in place to replace those roles. “Most of what’s happening is in anticipation of AI’s impact,” he said. “Not because the capabilities already exist.”

This phenomenon, sometimes referred to as “AI washing,” reflects a broader disconnect between hype and reality.

But while Davenport is concerned about the impact of AI on jobs, he cautioned against being overly pessimistic.

“It’s not particularly helpful to make these dire predictions until we actually know,” Davenport said. “I think there’s plenty of cause for concern, but how many jobs and how quickly? I don’t think anybody really knows.”

History offers perspective. Davenport pointed to research showing that even with the rise of ATMs and online banking, the number of bank tellers remained relatively stable for decades. Rather than eliminating jobs outright, technology often reshapes them.

“Companies are slow to change,” Davenport said. “The technology part generally tends to be the easiest. Redesigning the business processes, integrating AI with existing technology, and retraining employees, that takes quite a while.”

The AI Bubble

Davenport also sees signs of an “AI bubble,” though not necessarily a dramatic collapse.

“We’ve seen a decline in the stocks of some companies that are heavily investing in data centers and so on,” Davenport said. “I think we have started to see the beginnings of a deflation of the bubble. I hope that if there is a bubble, it will deflate gently.”

“Redesigning the business processes, integrating AI with existing technology, and retraining employees, that takes quite a while.”
Babson Professor Thomas Davenport

Davenport pointed to recent research indicating some investment was based on expectations of rapid growth in AI demand. “Companies that have already invested heavily in AI aren’t necessarily eager to invest much more,” he said.

At the same time, advances in smaller, more efficient AI models could reduce the need for massive computing resources, challenging some of the assumptions driving current valuations of AI and data center providers.

A Call for Caution

Beyond jobs and markets, Davenport raised broader concerns about AI’s societal impact, from reliability issues to ethical risks.

Current systems, he noted, still make basic errors, even when they appear confident. That lack of reliability may slow adoption in high-stakes applications, but it also underscores the need for oversight.

“We need regulation. We need government help for people who lose their jobs,” Davenport said. “And we’re doing nothing.”

Federal action appears unlikely in the short term, he added, leaving a patchwork of local regulations and increasing reliance on market forces and consumer sentiment to shape AI adoption.