AI can speed up drafting while adding review work. Examine reported fatigue associations and what leaders should count in productivity estimates.
On October 12, 2026, HRD published an analysis of decision fatigue in AI-assisted work. It described a shift in cognitive effort from generating content toward selecting and evaluating it. The concern was repeated quality judgments rather than the act of prompting alone.
HRD illustrated the issue with AI-suggested email replies and AI-generated meeting summaries. Workers still need to decide whether a reply sounds appropriate or which meeting points matter. The outlet framed that stream of small judgments as a potential source of decision fatigue.
For founders and executive teams, the productivity question is therefore broader than drafting speed. The relevant endpoint is work that has been checked and approved. Counting only the first draft leaves out the review effort that HRD describes.
HRD's account focuses on how work changes when a person evaluates generated material. Its phrase “from generation to selection” captures the distinction. Producing text and deciding whether that text is useful are different parts of a task.
In its September 2026 discussion, Boston Consulting Group examined the human role in AI-enabled work. BCG reported a comparison between workers with more-intensive responsibilities for monitoring and evaluating AI outputs and workers with less-intensive oversight responsibilities.
The reported outcomes were mental effort, mental fatigue and information overload. Workers with more-intensive oversight reported higher levels of all of them. This gives the discussion a measurable foundation beyond the everyday examples in HRD's article.
The World Economic Forum attributes the findings to a 2026 Harvard Business Review study of 1,488 full-time US workers. That attribution provides a population description for the comparison. The findings concern what workers reported about their experience of AI oversight.
The distinction between the sources matters. HRD offers an interpretation of repeated judgment in everyday work. BCG and WEF report comparisons involving oversight intensity and workers' reported experiences. Together, they put selection and verification inside the productivity discussion rather than treating those activities as incidental.
For ExecuFuel, the relevant leadership objective is sustained performance rather than headline output speed. The practical implication of these reports is to account for the people who check the work. Review belongs in capacity planning alongside generation.
This reframes the gap between faster output and business value. A faster draft is an intermediate result. The business question is how much effort is needed to reach an acceptable final version and who carries that effort.
The research supports looking at the full task: generation followed by checking, editing and approval. Leaders can assess those stages together instead of declaring a productivity gain at the moment text appears. That keeps the estimate connected to the work described by HRD.
HRD's email example shows why drafting speed and decision load need separate attention. A suggested reply can arrive quickly while still requiring a judgment about tone. In its meeting-summary example, the judgment is about relevance. Those decisions remain part of the task even when the initial text is generated automatically.
The practical question is not simply whether the tool saves drafting time. It is whether the complete workflow serves the team well once selection and review are included. The focus is the total demand of the task rather than the visible speed of its opening stage.
BCG argues that effective oversight depends on the reviewer's knowledge and judgment. Assigning a person to approve an output is not the same as establishing that the person can assess it. Leaders should consider the capability required by the work being reviewed.
This makes protecting human judgment in AI-assisted work an operating concern. Responsibility for approval should be considered alongside the ability to evaluate quality. The useful question is whether the reviewer understands the material well enough to make a meaningful decision.
The reports also support monitoring the amount of checking and the perceived effort involved. A team can examine whether AI-assisted tasks are creating more review work while separately tracking output and turnaround time. These are different aspects of adoption.
That perspective fits a broader discussion of setting realistic boundaries around team workload. Adding AI-assisted work to a process should prompt an accounting of review responsibilities. The aim is to make that work visible rather than leaving the people doing it outside the productivity calculation.
BCG reports that workers with more-intensive AI oversight responsibilities reported:
The comparison was with workers who had less-intensive oversight responsibilities. The World Economic Forum reports the same associations in its account of the Harvard Business Review study.
For executive teams, the significance is the spread of outcomes being considered. The discussion is not limited to how many drafts workers produce. It also includes the effort and fatigue they report while supervising the output.
These figures give leaders a reason to investigate the human side of their own workflows. Their practical value is directing attention to review effort and reported strain alongside claims of faster delivery.
The comparison is an association between oversight intensity and reported experience. It does not establish that AI oversight caused the differences. It also does not establish clinical burnout or show that the same burden exists across industries and organizations. The percentages should not be used as fixed adjustments to a productivity forecast.
BCG's page does not provide the study's sample size or population details. WEF identifies full-time US workers and supplies the sample count. BCG also omits measurement methods, instruments and statistical uncertainty estimates. It also leaves open how underlying job demands contributed to the differences.
The matching figures in BCG and WEF should not be counted as independent replications. HRD's analysis is also not a population-wide measurement of decision fatigue or the time consumed by review.
HRD acknowledges that AI can help address productivity pressures. Its argument is that review work can sit alongside that benefit. The evidence therefore supports assessing the complete workflow rather than assuming either that AI always improves productivity or that it always reduces it.
Finally, these sources motivate monitoring review work and perceived effort without validating a particular measurement method. Such monitoring is an operational inquiry rather than a health assessment. Personal health decisions belong with qualified care rather than conclusions drawn from workplace comparisons.
The next useful research would follow AI-assisted work from generation through final approval. It should examine review effort alongside task completion and quality. Following workers over time would also help distinguish a sustained pattern from differences captured at a particular point.
Another important question is how reviewer capability relates to oversight demands. BCG's emphasis on knowledge and judgment makes that a priority for further investigation. Research should distinguish the amount of review assigned from the person's ability to perform it well.
For the industry, a useful shift would be to treat review capacity as a design requirement. Productivity assessments would then account for the human work between an available draft and an accepted result. That would bring adoption claims closer to the work leaders are actually responsible for organizing.
The practical conclusion is not to reject AI or disregard the speed it can bring to generation. It is to make selection and verification visible. Stronger workplace evidence should clarify when AI reduces total effort and when it redistributes that effort toward human judgment.
Stay connected for research and practical guidance on executive performance, energy, focus, sleep, recovery and longevity. Ideas built for people who want to stay sharp, capable and effective for the long run.
Build habits and systems that support clear thinking, steady energy and long term capacity throughout a demanding career.
explore the Blog
