How much further could organizations go if they fully leveraged the human capacity gained when AI takes on mundane, repetitive tasks?

Artificial intelligence has quickly become one of the most definitive business conversations of our time. As organizations invest billions into generative AI, automation, and other technologies that promise to improve productivity, reduce costs, and eliminate repetitive tasks, much of the discussion remains focused on how much time and money AI can save, how much faster people can work, and which jobs or functions may be transformed (or, sadly, eliminated) in the process.

Those are worthwhile questions, and the evidence suggests that AI can indeed deliver meaningful productivity gains under the right conditions. For example, Brynjolfsson et al. (2023) found that “AI assistance increases worker productivity, resulting in a 14% increase in the number of chats that an agent successfully resolves per hour” (p. 2).

Studies have consistently found improvements in speed when AI is used appropriately. In their 2022 controlled experiment on the impact of AI on developer productivity, Peng et al. (2023) measured the productivity impact of using GitHub Copilot in programming tasks among 95 professional programmers with an average of six years experience. Participants were instructed to write an HTTP server in JavaScript, where the treatment group used GitHub Copilot to complete the task, and the control group did not. They highlighted the following findings:

"[Researchers] calculated two metrics as a measure of performance for each group: task success and task completion time. The performance difference between treated and control groups are statistically and practically significant: the treated group completed the task 55.8% faster (95% confidence interval: 21-89%). Developers with less programming experience, older programmers, and those who program more hours per day benefited the most.

Finally, this study does not examine the effects of AI on code quality. AI assistance can increase code quality if it suggests code better than the programmer writes, or it can reduce quality if the programmer pays less attention to code. The code quality can have performance and security considerations that can change the real-world impact of AI.”

While I find these studies to be significant, believe they are also incomplete. They focus almost entirely on the technology while overlooking a far more important organizational question.

What happens to the human capacity that AI creates?

That question shifts the conversation away from technology and toward organizational behavior. AI may create opportunities to reclaim time, reduce routine work, and increase efficiency, but organizations ultimately decide what happens next. Some will reinvest that capacity in developing people, strengthening collaboration, improving decision-making, and building organizational capability. Others will quietly absorb every reclaimed hour into additional work, tighter deadlines, and higher productivity targets. The technology may be identical, but the organizational choices are not. This is where AI becomes far more interesting than another productivity tool. It becomes a mirror.

AI Is the Mirror That Reveals Organizations to Themselves

Every major disruption reveals something about an organization. Economic downturns expose financial resilience (or lack thereof). Mergers reveal cultural compatibility. In 2020, as remote work exploded, it exposed levels of trust (or its absence), autonomy, and leadership capability that many organizations had never examined closely. AI is doing much the same thing currently. It’s revealing what organizations truly value when they are given choices.

Organizations often describe themselves through their mission statements, values, and leadership messages— making public attestations of being innovative, people-centered, learning-oriented, or committed to developing talent. Those declarations certainly matter, but they are only part of the story. Organizations ultimately reveal their priorities through the choices they make every day. They reveal them through where time is invested, where money is allocated, what gets measured, what gets rewarded, what behavior is tolerated, and which initiatives are repeatedly postponed. AI simply provides another opportunity to observe those choices.

When technology creates additional human capacity, leaders must decide whether to reinvest it in learning, innovation, collaboration, and organizational improvement or absorb it into more output. Those decisions reveal far more about an organization than any carefully crafted statement hanging on the wall.

Economists have long described this concept as revealed preferences. Organizations often communicate one set of priorities while allocating resources in ways that demonstrate another. AI is making those differences increasingly difficult to hide.

The Capacity Dividend May Be Larger Than the Productivity Dividend

As already mentioned, research consistently demonstrates that generative AI can improve performance on many routine tasks. These findings are important because they establish that AI can create what might be called a capacity dividend. People recover time that was previously consumed by repetitive or administrative activities. However, organizations often falsely assume that this recovered capacity automatically translates into higher-value work. History suggests otherwise.

Organizations have spent decades improving efficiency through technology. Email replaced paper correspondence. Enterprise software streamlined administrative processes. Collaboration platforms reduced communication barriers. Automation simplified countless operational tasks. Yet relatively few people would argue that organizational life has become less busy. Instead, every efficiency gain has often been accompanied by new expectations, additional responsibilities, more reporting requirements, increased responsiveness, and greater workloads.

In many organizations, work expands to consume every efficiency that technology creates. This is the organizational version of Parkinson’s Law: work expands to fill the capacity created by efficiency. That pattern should concern leaders because it reflects a fundamental difference between improving productivity and strengthening organizational capability. Where productivity measures how efficiently work is completed, capability determines whether an organization can continue learning, adapting, solving increasingly complex problems, and executing its strategy over time. Those outcomes require something more than speed alone can deliver.

Capability Does Not Build Itself

One of AI's greatest strengths is its ability to provide performance support in the flow of work. It can summarize information, suggest solutions, draft documents, answer routine questions, and make expert knowledge more accessible to less experienced employees. Properly implemented, these capabilities can accelerate learning and improve performance. They can also create unintended consequences if organizations confuse assistance with development.

Professional expertise develops through experience, observation, practice, feedback, reflection, and progressively more challenging work and it rarely emerges in a single leap. Many of the routine tasks now being automated have historically served as part of that developmental journey. Analysts grow from Junior to Senior status by learning how to conduct research. New consultants develop judgment through documentation and analysis. Early-career professionals build pattern recognition by repeatedly working through foundational tasks.

If organizations automate these expertise-developing experiences without intention, they risk shortening the apprenticeship runway while expecting the same level of professional judgment at the end. This is not an argument against AI. It is an argument for organizational design. Technology can support capability development, but it cannot replace the leadership decisions required to build it.

Organizational Capability Is Built Through Choices

For quite some time, it’s been recognized that performance problems rarely originate from individual effort alone. They emerge from the interaction between people, processes, systems, incentives, leadership, resources, and organizational culture. AI simply enters that existing system. Imagine two organizations implementing exactly the same AI platform.

• “Company A” uses the recovered capacity to strengthen coaching, improve cross-functional collaboration, redesign inefficient processes, encourage experimentation, and create opportunities for continuous learning.

• “Company B” increases utilization targets, adds additional reporting requirements, reduces staffing, and expects employees to produce more within the same amount of time.

Both organizations may report productivity improvements but only one is intentionally strengthening organizational capability. The difference has very little to do with technology and has everything to do with leadership and organizational choices.

The Leadership Question That Should Be Being Asked

Perhaps the most important question surrounding AI is not whether it will change work. That answer is becoming increasingly obvious. The more important question is what leaders choose to do when work changes. Will reclaimed capacity become an opportunity for deeper thinking, stronger collaboration, better coaching, more innovation, and improved organizational learning? Or will it quietly disappear beneath the weight of additional expectations?

Organizations often speak passionately about resilience, adaptability, innovation, and continuous improvement. However, those capabilities don’t emerge simply because leaders declare them important. They require time to think and plan, permission to allow opportunities to learn, psychological safety to experiment, and systems that reward improvement rather than constant activity. I assert that—most importantly—what’s required now, more than ever, is the patience and courage to think differently and make better decisions about what’s important.  

If every efficiency gained through AI is immediately consumed by additional work, organizations may become faster without becoming any more capable. That may prove to be one of the greatest missed opportunities of this technological era.

Looking Beyond the Technology

AI is giving organizations one of the most significant leadership choices they have faced in decades. Some will use it primarily to accelerate work. Others will use it to strengthen the capabilities that determine whether they can continue adapting, learning, and performing in increasingly complex environments. The technology itself will not determine which path an organization follows. Leadership decisions and action (or inaction) will do that.

Therefore, the conversation cannot end with software selection, prompt engineering, or productivity targets. Ultimately, in order to thrive, organizations need a deeper understanding of how work is designed, how people learn and develop expertise, how capability is built across interconnected systems, and how leadership decisions shape long-term performance. Those are organizational capability questions that demand carefully thought out appropriate solutions.

---

References

Brynjolfsson, E., Li, D., & Raymond, L. R. (2023). Generative AI at work (NBER Working Paper No. 31161). National Bureau of Economic Research. https://doi.org/10.3386/w31161

Peng, S., Kalliamvakou, E., Cihon, P., & Demirer, M. (2023). The impact of AI on developer productivity: Evidence from GitHub Copilot. arXiv. https://doi.org/10.48550/arXiv.2302.06590