Can Artificial Intelligence Increase Productivity?

The Productivity Promise—and the Puzzle Behind It

A curious paradox sits at the center of modern economic debate. Every few decades, a new technology arrives accompanied by extraordinary predictions. Factories will become more efficient. Offices will become leaner. Workers will accomplish in hours what once required days. Economic growth, we are told, will accelerate.

Artificial intelligence is the latest entrant in this long tradition of technological optimism. Its advocates argue that algorithms capable of writing reports, generating software code, analyzing medical images, and answering customer inquiries represent the most consequential productivity-enhancing innovation since the computer revolution.

Yet history encourages caution.

The mere existence of a powerful technology does not guarantee broad-based productivity growth. Economies are littered with examples of remarkable inventions whose effects fell short of expectations. Some technologies transformed production and living standards. Others generated excitement without fundamentally altering economic performance.

The question, therefore, is not whether artificial intelligence is impressive. It plainly is. The more consequential question is whether AI can increase productivity on a meaningful scale—and if so, under what conditions.

The answer is yes. But the path is considerably more complicated than many headlines suggest.

Understanding Productivity: More Than Working Faster

When economists discuss productivity, they are not simply referring to people working harder or longer hours. Productivity measures how efficiently inputs such as labor, capital, and knowledge are converted into output.

At its most basic level, productivity growth means producing more with the same resources.

A restaurant that serves twice as many customers with the same staff has become more productive.

A manufacturer that produces identical goods using fewer labor hours has become more productive.

A software company whose engineers develop applications in half the previous time has become more productive.

Historically, sustained increases in living standards have depended heavily on productivity growth. Wage growth, business profitability, and long-term economic expansion all rely on societies becoming better at generating value from existing resources.

This is where artificial intelligence enters the discussion.

AI promises not merely to automate tasks but to augment human capabilities. That distinction matters.

Why AI Differs from Previous Waves of Automation

Many earlier technologies focused on replacing routine physical tasks.

Industrial machinery reduced the need for manual labor in manufacturing. Automated assembly lines standardized production. Robotics eliminated repetitive actions on factory floors.

Artificial intelligence operates differently.

Instead of targeting physical effort, AI increasingly targets cognitive activities: drafting documents, summarizing information, diagnosing problems, generating designs, and assisting decision-making.

This shift expands the range of occupations potentially affected.

Lawyers can use AI to review contracts.

Doctors can use it to identify anomalies in medical scans.

Marketers can generate campaign drafts.

Programmers can accelerate software development.

Teachers can prepare lesson plans.

Researchers can synthesize large bodies of literature.

In each case, the technology is not necessarily replacing the worker. Rather, it is changing the set of tasks that worker performs.

The distinction between task replacement and task augmentation is crucial because productivity gains emerge differently under each scenario.

The Evidence So Far

Early empirical evidence suggests that AI can generate substantial productivity improvements in specific contexts.

Several studies have found that workers equipped with generative AI tools complete tasks more quickly while maintaining comparable levels of quality. The largest gains often appear among less experienced employees, who benefit from AI-generated guidance and recommendations.

Consider customer service operations. AI systems can instantly retrieve relevant information, draft responses, and assist representatives during conversations. Instead of searching through extensive documentation, workers receive suggestions in real time.

Similarly, software developers increasingly rely on AI coding assistants to generate boilerplate code, identify errors, and propose solutions. What once required hours may now require minutes.

Yet these successes should not be confused with economy-wide productivity transformation.

History teaches an important lesson: local productivity gains do not automatically translate into aggregate economic growth.

A Comparison of AI's Potential Productivity Effects

The table illustrates an important point. Productivity gains depend not merely on technological capability but also on organizational adaptation.

That adaptation is often the most difficult part.

The Missing Ingredient: Organizational Change

One lesson repeatedly emerges from economic history.

Technological breakthroughs rarely deliver their full benefits immediately.

The introduction of electricity provides a classic example. Early factories simply replaced steam engines with electric motors. Productivity improvements remained modest. Only later, when firms reorganized production processes around the flexibility electricity enabled, did substantial gains materialize.

Artificial intelligence may follow a similar trajectory.

Many firms currently deploy AI as an add-on tool rather than redesigning workflows around its capabilities. Employees use chatbots occasionally. Managers experiment with automated reporting. Departments test pilot programs.

These incremental changes can produce benefits, but transformative productivity growth requires deeper institutional adjustment.

Companies must rethink job design.

Training systems must evolve.

Management practices must adapt.

Workflows must be reorganized.

Without these complementary investments, AI's impact may remain surprisingly limited.

A Lesson I Learned Watching AI Adoption

Several months ago, I observed two teams within the same organization experimenting with identical AI tools.

The first team integrated AI into virtually every stage of its workflow. Employees revised procedures, established verification protocols, and redesigned task allocation. Within weeks, project completion times fell dramatically.

The second team approached AI as a convenience. Workers occasionally used it for drafting emails or generating summaries, but existing processes remained largely unchanged.

The results were striking.

Both groups possessed access to the same technology. Yet only one experienced meaningful productivity gains.

The lesson was straightforward: technology alone rarely creates transformation. Productivity improvements emerge when organizations alter the way work is structured.

Economists have encountered this pattern repeatedly across centuries of technological change.

AI appears unlikely to be an exception.

Why Some Economists Remain Skeptical

Despite the enthusiasm surrounding artificial intelligence, skepticism persists among economists for good reason.

Productivity growth across advanced economies has been relatively sluggish for much of the past two decades. This slowdown occurred despite extraordinary advances in computing power, internet infrastructure, and digital technologies.

Why should AI be different?

One concern is that many AI applications focus on activities that generate convenience rather than substantial economic value.

If employees spend less time drafting emails but continue attending the same meetings and navigating the same bureaucratic processes, aggregate productivity gains may remain modest.

Another concern involves measurement.

Economic statistics often struggle to capture improvements in quality. If AI allows workers to produce better outputs rather than simply more outputs, official productivity data may underestimate its impact.

The challenge, therefore, lies not only in generating productivity gains but also in identifying where those gains truly emerge.

The Distribution Question

Perhaps the most important issue concerns who benefits from productivity growth.

Historically, technological progress has not automatically improved outcomes for all workers.

Some innovations increased demand for skilled labor while reducing opportunities for others.

Artificial intelligence could generate similar dynamics.

If AI primarily augments highly educated professionals, productivity gains may become concentrated among a relatively small segment of the workforce.

Conversely, if AI helps less experienced workers perform complex tasks more effectively, it could broaden access to valuable opportunities and reduce skill disparities.

The distinction matters enormously.

A technology that increases productivity while widening inequality creates a different economic landscape from one that increases productivity while expanding opportunity.

The future impact of AI depends not merely on technical development but also on choices made by businesses, policymakers, and institutions.

What Would a True Productivity Revolution Look Like?

A genuine AI-driven productivity revolution would exhibit several characteristics.

First, productivity gains would appear across multiple sectors rather than remaining confined to a handful of industries.

Second, firms would redesign workflows around AI capabilities instead of simply layering AI onto existing processes.

Third, workers would receive training that enables effective collaboration with intelligent systems.

Fourth, innovation would accelerate as AI assists research, discovery, and problem-solving activities.

Most importantly, the technology would complement human capabilities rather than merely replacing them.

The largest gains in economic history often emerged when technology expanded what workers could accomplish, not when it simply reduced labor costs.

That observation may prove especially relevant today.

The Real Opportunity

The most consequential productivity effects of artificial intelligence may not stem from automating existing tasks.

They may arise from enabling entirely new forms of production.

The internet did not merely make communication faster. It created industries that scarcely existed before.

Likewise, AI may generate new products, services, and business models that are difficult to envision today.

Some of the most important economic benefits could emerge from innovations that remain invisible at present.

This possibility complicates forecasts. It also explains why both excessive optimism and excessive pessimism are misguided.

Technological revolutions rarely unfold according to simple narratives.

Conclusion: Productivity Is a Choice, Not a Destiny

Artificial intelligence can increase productivity. The evidence already points in that direction.

But productivity growth is not embedded in the technology itself. It emerges from the interaction between technology, institutions, organizations, and workers.

That is the critical distinction often overlooked in public debate.

The question is not whether AI possesses extraordinary capabilities. It clearly does.

The question is whether businesses and societies will redesign systems to make effective use of those capabilities.

History offers a sobering reminder. Many transformative technologies spent years—or even decades—underperforming expectations before complementary changes unlocked their potential.

Artificial intelligence may ultimately become one of the most productive technologies ever created. It may also disappoint those expecting immediate economic transformation.

The outcome depends less on algorithms than on choices. Productivity is not an automatic consequence of innovation. It is the result of how innovation is deployed, governed, and integrated into economic life.

And that is why the future of AI remains, above all, a question of institutions rather than machines.