We are moving into the AI era before we have settled the bill for the smartphone era. This may explain why some of the pushback against AI is less about robots or job losses and more about a sense of repetition. The promises are familiar: save time, remove routine, connect everyone, create more freedom. On paper, much of this happened. We can reach anyone, work from anywhere, and access information instantly. We can also receive seventeen urgent messages before breakfast. Progress delivered what it promised. It just left out the part about operating costs.
The invoice from the last technology wave is still open. It appears as permanent availability, fragmented attention, and the expectation that people can monitor multiple channels, join meetings, and still deliver focused work. We call this productivity, mostly because ‘continuous cognitive assault’ does not fit in a quarterly report. Now AI arrives with a new offer: it can write, summarize, analyze, generate options, and clear hours of routine work. The real question is what organizations will actually do with those saved hours.
The optimistic version is that people will use the freed-up time for deeper thinking, creativity, and recovery. The more common pattern is that saved capacity becomes the new baseline. If someone can now produce twice as much, the organization expects twice as much, usually by next quarter and ideally without a bigger budget. This is how efficiency tends to work in real systems. A tool saves thirty minutes, so another task appears. Automation removes a process, so management adds three new reporting requirements. AI drafts something in seconds, but someone still needs to review, verify, adjust, compare, and take responsibility when the confident output turns out to be wrong.
Work rarely disappears. It shifts from creation to supervision, and supervision can be just as draining. Take a product manager before AI: review research, prepare a proposal, discuss with the team, make a decision, explain it to stakeholders. Now add five AI tools: one for customer feedback, one for roadmaps, another for requirements, another for competitor analysis, and one more for meeting notes. On paper, this looks like a workload reduction. In practice, the manager now inspects five streams of output, spots contradictions, questions assumptions, checks sources, decides what matters, and explains why the AI-generated strategy is not automatically the company strategy. The job has not shrunk. It has become a control tower, and the planes keep coming.
This is where the conversation about technology gets selective. We measure what machines produce, but rarely what humans have to absorb. Drafts created, tickets resolved, reports generated, minutes saved: all easy to count. The extra verification, monitoring, switching, interpretation, and decision-making? Not so much. The system sees output. The nervous system sees Tuesday.
This pattern is not new. Social networks promised connection and monetized attention. Smartphones promised mobility and turned every location into a potential workplace. Collaboration platforms promised transparency and delivered environments where key information hides in emails, chats, boards, comments, or meetings that could have been emails but, in a strategic twist, became both. The tools are not the problem. The issue is that every benefit creates new costs, and those costs rarely land in the same place as the benefits.
Companies get growth, data, speed, scale, and higher output. Individuals pay with attention, recovery, sleep, and the ability to stay with a hard problem long enough to solve it. When people struggle, organizations offer help: time-management training, resilience workshops, meditation apps, and a digital wellbeing webinar scheduled during lunch because calendars are full. The message is polite but clear: the system is fine; the employee needs an upgrade.
We keep repairing people and returning them to the same environment that exhausted them. Personal habits do matter. People can disable notifications, protect focus time, improve boundaries, and use technology more consciously. But individual discipline cannot fully compensate for a system designed around constant responsiveness, excessive information, and permanent task switching. You can practise digital hygiene. You cannot personally redesign your company’s operating model.
This is why cognitive overload should not be discussed only as an individual weakness. In many workplaces, it is an expected result of how work is structured. If a role requires someone to monitor several platforms, react continuously, make decisions under time pressure, and switch between strategic and operational questions every few minutes, exhaustion is not surprising. It is not a mysterious failure of resilience. It is the output of the system.
The uncomfortable reality is that companies have little incentive to look too closely. Organizations measure what supports investment decisions: revenue, engagement, adoption, throughput, utilization, conversion, growth. The cost to human attention is harder to quantify and less convenient to discuss. A company rolling out AI wants to show time saved, not that the saved time is now filled with more supervision, more parallel work, and higher expectations. A platform wants to show engagement, not whether users can still read ten pages without checking their phones.
The side creating the effect rarely wants to measure the harm. The side experiencing it rarely has the data or power to do so. The cost disappears into the gap between what is profitable and what is measurable.
This is not an argument against AI. AI may be one of the most valuable tools in decades. It can remove meaningless work, expand access to expertise, improve decisions, and give small teams capabilities that used to require large organizations. But none of this is automatic. AI can reduce cognitive load, or multiply it. It can protect time for deeper work, or create more streams to monitor. It can give people more autonomy, or just increase the volume they are expected to handle.
Technology does not make that decision. Management does.
The real question is not whether AI makes people more productive. It is what happens to the capacity AI creates. Does the organization return some of it as time to think, recover, learn, and improve decisions? Or does it immediately convert every saved minute into another task, another target, and another dashboard to prove the transformation worked?
Progress always comes with a price. Industrial growth cost bodies, cities, and safety before rules caught up. The digital economy cost attention, privacy, boundaries, and recovery. AI will have a price too. The goal is not to stop progress or pretend the past was better. It is to stop treating the human cost as a personal problem discovered after the rollout.
Responsible progress means measuring not just what technology lets people produce, but what people have to carry to produce it. When every new tool promises to save time, but nobody has any time left, maybe the tool is not the only part of the system that needs a closer look.