AI Doesn't Just Automate Tasks: It Changes Who You Are at Work
Artificial intelligence is producing a contradictory effect that few organizations measure: it makes work more satisfying and more demanding at the same time. The BCG AI at Work 2026 study calls this the joy paradox: most regular users enjoy their work more since using AI, but a large share of them report an increased mental load.
The reason is that AI eliminates friction but adds new responsibilities: reviewing results, detecting errors, correcting hallucinations, and deciding when to trust the machine. The worker stops being an executor and becomes a supervisor. And not everyone adapts to that change in the same way.
What the Data Says: The Executor Becomes a Supervisor
The change is not marginal. According to BCG, most frontline employees already use AI regularly, and among managers and leaders the use is almost universal. AI has moved beyond technical layers and become transversal.
This massive use reconfigures job architecture. Most respondents believe that the skill expectations for their role have changed, a majority state that AI has taken over simple tasks, leaving them with more complex work, and almost half say their role has shifted towards managing and directing artificial intelligence.
Microsoft, in its Work Trend Index 2025, names this new role: the agent boss, the employee who directs AI agents as they previously coordinated with people. But supervising a machine—deciding when to trust it and when not to—is a cognitively and emotionally different job from executing. Focus Values System insight: Not all axiological profiles thrive equally in a supervisor role. The Hartman Value Profile (HVP) from Focus Values System measures judgment capacity under pressure, tolerance for ambiguity, and the discernment to validate or reject a result—exactly the capabilities that distinguish a good agent boss—before assigning someone to that role.
Why Mental Load Is Not Distributed Equally
The share of people reporting an increased mental load hides an uncomfortable truth: this load does not fall uniformly across the workforce. It depends on how each person values work and processes responsibility.
- The profile that needs structural certainty: experiences supervising a fallible AI as a constant source of stress: they never know if the result is reliable.
- The detail-oriented and quality control profile: absorbs the verification task well but can become overwhelmed trying to review everything.
- The profile activated by autonomy: enjoys the liberated time but gets frustrated if the organization doesn't provide direction on how to use it.
- The people-oriented profile: misses the human component when part of their interaction is delegated to agents.
The same tool, the same productivity figure, and four radically different internal experiences. Treating mental load as an average means losing sight of those who are approaching exhaustion.
The Trap of Liberated Time
There's a second piece of data that should be read alongside the mental load figure. Many frontline employees using AI save at least one workday per week. But most of them receive little to no guidance on how to reinvest that time.
Therein lies the real problem. The decisive question is no longer how much time AI saves, but who decides what that new capacity is dedicated to. When that decision is left to chance, liberated time does not convert into higher-value work: it fills with more supervision, more review, and more meetings. The paradox worsens. Focus Values System insight: The value profile indicates which more demanding tasks activate each person's best performance and which ones wear them down. This information allows for deliberately reinvesting liberated time—directing each person towards work that provides them with strength—instead of letting it fill with indiscriminate load.
Efficiency and Wellbeing Are Not Opposed (If There Is Discernment)
The most useful finding of the study dismantles a common dichotomy: business efficiency versus employee wellbeing. BCG argues that companies that capture the most value with AI are also those where employees report greater satisfaction.
The condition is that five things exist: strategic clarity, measurement, consistency between discourse and action, employee participation in work redesign, and recognition of adoption. Where this structure is lacking, AI only adds load. Where it exists, it adds meaning.
The most significant factor, according to BCG, is not access to tools but clarity of direction: "employees with clear strategy but limited access to tools report more impact than those with many tools but little direction." Buying software does not transform an organization; knowing which human capabilities need to become more valuable does.
From Paradox to Plan: How to Act
Transforming the joy paradox from a risk to an advantage requires treating the human factor with the same rigor as the technical one. For an HR department, the path has four steps:
- Map who supervises what: identify which individuals are taking on agent boss roles, formally or de facto.
- Measure axiological resilience: understand which profiles tolerate ambiguity and constant verification well, and which need more certainty or support.
- Distribute cognitive load with discernment: assign supervision according to profile, not just hierarchy or availability.
- Personalize upskilling: the skills gap is real—almost everyone believes they will need relevant training in the coming years, but only a minority feel prepared—and generic courses do not close it.
The zero step in all of this is knowing people. Not by intuition, but with data.
In summary:
What is the "joy paradox" of AI at work?
It is the finding from the BCG AI at Work 2026 study that artificial intelligence makes work more satisfying and more demanding at the same time: most regular users enjoy their work more since using AI, but a large share of them report an increased mental load. The cause is that AI eliminates routine tasks but adds the responsibility of reviewing, verifying, and deciding when to trust the machine.
What is an "agent boss"?
It is the term Microsoft (Work Trend Index 2025) uses for the employee whose work shifts from executing tasks to directing and supervising AI agents. According to BCG, almost half of professionals already state that their role has moved towards managing artificial intelligence. It is a cognitively distinct role from that of an executor, and not everyone assumes it with the same ease.
How does the HVP help manage AI's mental load?
The Hartman Value Profile (HVP) measures how each person values and prioritizes, what type of work activates their best performance, and what wears them down. This allows anticipating who adapts well to the AI supervisor role, distributing the verification load according to resilience, and personalizing training, protecting wellbeing while the organization captures the value of technology.
Conclusion
AI is already changing not only what we do, but who we are at work: from executors to supervisors, from operators to agent bosses. This change brings more satisfaction and more load at the same time, and the difference between one and the other depends on the specific individuals experiencing it.
Treating this load as an average is the mistake. The competitive advantage is not in having more tools, but in knowing who each person is and what they need to thrive in the new scheme. That is the organizational intelligence the study calls for.
The HVP from Focus Values System provides the human data that work redesign needs: who adapts to the load of the new role, who is activated by autonomy, and what each person needs to continue contributing value when the machine executes.
Source: "AI is not enough: organizational intelligence will redesign your work." Expansión&Empleo (Editorial, Madrid), June 4, 2026. Base study: BCG "AI at Work 2026" ("Strategy matters more than tools"). Sources cited in the original article: McKinsey ("The State of AI"), World Economic Forum (Future of Jobs Report 2025), Microsoft (Work Trend Index 2025), Accenture (Talent Reinventors) and Gartner.