Productivity’s AI Promise Is Murkier Than CEOs Admit

The tone on recent earnings calls has been one of confidence. Leaning forward and speaking steadily, executives describe how AI systems are “unlocking efficiencies” and “streamlining workflows.” Glass towers in San Francisco and conference rooms in Midtown Manhattan both have slides flashing across screens. Something historic seems to be happening. However, the statistics seem more subdued than the rhetoric.

According to a National Bureau of Economic Research survey, over 80% of 6,000 executives said AI had no appreciable effect on productivity or employment. That figure stands in stark contrast to the optimism that permeates boardrooms. Businesses claim to be widely implementing AI. Over the next three years, many anticipate increasing usage. But quantifiable improvements? Still elusive. Perhaps the discrepancy isn’t exactly dishonesty. Perhaps anticipation is stronger than proof.

The future becomes real when you walk through a logistics warehouse that is experimenting with AI-driven robotics. Under the guidance of ever-more-advanced vision systems, machines move between shelves, lifting crates with mechanical arms that mimic simplified human limbs. Engineers congregate around laptops, adjusting battery life, motion sequences, and algorithms. After previous software investments, money is now pouring into this hardware wave. Robotics venture funding has increased dramatically in recent years, and valuations have increased along with it.

However, productivity isn’t a demonstration. This obstinate macroeconomic figure is influenced by millions of daily tasks, such as assembling parts, responding to emails, and updating hospital charts at three in the morning. It is difficult to gauge progress across that sprawl. According to Stanford’s Erika McEntarfer, it might be very difficult to separate AI’s impact from other factors like changes in the labor market and new capital investment.

Investors appear to think the profits will materialize anyhow. Businesses that make compelling claims about automation increasing margins are rewarded by markets. There is a subtle financial logic at work here: more predictable earnings are promised by substituting depreciable capital assets, such as robots, for variable labor costs. When that story is told frequently enough, confidence is bolstered even before the savings become apparent.

History warns us. Economists were perplexed by the so-called information technology productivity paradox in the late 1990s. Despite the widespread presence of computers, worker output hardly changed. It was only after complementary systems and skills caught up that the gains became evident in the data. AI might be at a similar turning point, strong in theory but awaiting organizational reform to achieve true efficiency. Still, there’s a sense of déjà vu as you watch the current cycle play out.

In a German manufacturing facility where regional pride was once defined by automotive precision, engineers are now applying their knowledge to robotic actuators and motion systems. A sizable portion of the supply chain for high-precision components is located in Europe. The same discipline that once produced engines and transmissions now permeates factories. The goal is to fill demographic gaps by having humanoid machines work alongside aging human workforces.

The demographic argument is strong. The population is aging quickly. Manufacturing, healthcare, and agriculture are all experiencing growing labor shortages. Humanoid robots could relieve that stress because they can navigate human-made environments. Global increases in defense spending are driving up demand for production automation.

However, according to the NBER survey, the majority of executives have not yet seen an increase in productivity. The moment is defined by this tension between structural necessity and immediate impact.

It’s possible that a portion of the issue is cultural rather than technical. When AI systems are implemented, previously concealed inefficiencies are frequently exposed. It is necessary to redesign processes. Data was cleaned. Redefining roles. The technology reveals the vulnerability of outdated workflows rather than just fitting them into place. If productivity increases occur, they might come after disruption rather than before.

The human element is another. Employees adjust differently. Some people use AI tools to analyze data more thoroughly or to draft reports more quickly. Others are hesitant because they are unsure of their dependability or job security. It’s still unclear if broad adoption will increase skill levels or covertly deskill segments of the workforce, causing value to shift upward and middle layers to disappear.

During lunch, workers outside a London office tower spill onto the sidewalk, some using AI chat tools to condense documents before afternoon meetings. The scene appears to be efficient and modern. Does it, however, produce quantifiable results? Or does it just make work feel different?

It’s difficult to ignore how heavily the AI narrative is predicated on expectations. For instance, humanoid robotics projections indicate that within ten years, the market will grow from a few billion dollars to tens or even hundreds of billions. The price of lithium-ion batteries has significantly decreased. Robots can now function in less structured environments thanks to developments in machine perception. Production costs are falling as the “brains, brawn, and batteries” all work together to improve.

However, increasing overall productivity and scaling a technology are not the same thing. Revenues from one can increase while those from the other stay the same.

CEOs seem to be overseeing two audiences simultaneously. They are experimenting, learning, and absorbing expenses on the inside. They are expressing confidence to the outside world by promising investors that they won’t be left behind by the AI revolution. It is rarely rewarding to admit uncertainty during a quarterly call.

Thus, the promise is still there—strong, reasonable, and even required in light of economic and demographic forces. In aging societies, the idea of having fewer workers and doing more work in less time is alluring.

Another question is whether and how quickly AI will produce that result at scale. The machines are getting better. Money has been committed. Executives are hopeful.As of right now, productivity figures are still catching up.