Artificial intelligence can reduce tasks that once took hours to just minutes, but the time freed up is often used for additional work rather than leisure. This contradiction appears both in international research and in the experiences of practitioners.
Research data and figures
In Boston Consulting Group's (BCG) 2025 survey of 7,492 workplace AI users, 54 percent of respondents said they spent the time saved using AI on additional tasks. Only 27 percent indicated they used the freed time for non-work activities. Forty-seven percent reported at least six hours of weekly time savings, yet roughly one-third received guidance on how to allocate that extra capacity.
BCG's follow-up in 2026 covered 14 markets and 11,749 workers: among regular AI users, 41 percent reported increased mental workload while 67 percent experienced improved workplace satisfaction.
Gallup's 2026 US survey found that 31 percent of AI-using employees reported receiving more tasks or responsibility because of the technology; among those who experienced faster work, the share was 37 percent.
Researchers at the University of California Berkeley's Haas School of Business observed a similar pattern. Studying a roughly 200-person US tech firm for eight months, they found AI users worked faster, handled a greater variety of tasks, and extended their work across longer periods of the day; the extra tasks were often self-initiated.
Domestic examples: faster processes, more initiatives
Tamás Tóth, founding expert of AI Workshop with three decades of software development and company leadership experience, provides practical illustrations. His company built a system that records client data from a phone call, generates a summary of next steps, and assembles an offer from the product database — reducing the time to prepare an offer after a call to three minutes, compared with previously much longer processes.
The technology can also help tackle long-postponed administrative work: with appropriate access and instructions, the system can scan, rename and organize the contents of up to 18,000 disordered files to reduce administrative burden.
The joy of creation and the risk of overload
Tóth argues that the ability to implement ideas that once seemed impractical creates strong momentum, which easily leads to multiple parallel projects. He calls this dynamic "AI-creation addiction": quick successes prompt immediate experimentation and a stream of new ideas, which can end up producing too many concurrent initiatives. In his own firm, so many prototypes were launched at once that some closing steps were missed.
This momentum can increase mental load: BCG's 2026 study measured a 41 percent rise in reported mental strain among regular AI users.
Management and work organisation: deciding when enough is enough
Tóth sees it as a leader's responsibility to decide when to stop and to close tasks. He checks the status of initiated work daily, groups multi-day tasks into projects for later follow-up, and limits the number of active ideas — for example, setting a cap of twelve active ideas enforced by his AI assistants.
He also stresses the importance of human oversight, careful control of access to company data, and measuring the business impact of each new development. A flashy prototype alone does not automatically produce revenue or cost savings.
The Hungarian context and implications
Rolling out AI in Hungary faces additional obstacles: Eurostat's 2025 data show that only 10.4 percent of Hungarian firms with at least ten employees used some form of AI, compared with a 20 percent EU average. As a result, Hungarian companies must contend with both implementing AI and learning how to use the capacity it frees in a sensible way.
Conclusion
AI can accelerate an increasing array of tasks and offers real opportunities to boost business efficiency. Yet evidence and practice indicate that time savings do not automatically translate into shorter workdays or higher profits; they often translate into more or different work. A key task for managers is therefore to pick AI initiatives that deliver measurable business value and to establish clear rules for handling parallel projects and workloads so that technological efficiency does not simply extend working time.
