AI is simplifying office workers’ document writing and coding, but a survey found it may take more time to review and edit the output. Some say adopting AI does not necessarily lead to productivity gains across an organisation.

Tech outlet TechRadar reported on Oct. 19 that AI services company Use.AI surveyed 8,316 adults using workplace AI across the United States, Britain, the European Union, Canada, Australia and Central and South America. It found 62 percent of respondents said AI made it easier to produce documents, presentation materials and code that they previously did not attempt because it required too much time and effort.

Problems emerged when AI-generated output was shared with others. Some 34 percent said they had passed on work written or assisted by AI without reading it to the end. Another 48 percent said they had received AI output that needed edits, and 43 percent said they had to remake inappropriate output from scratch.

In addition, 39 percent said they had experiences where reviewing AI output took longer than doing the work themselves. Another 41 percent said the time spent reviewing or verifying content increased after AI was introduced.

Some 37 percent also said problems with AI output delayed decision-making. It suggests that while time spent drafting is reduced, the burden of fixing errors can be shifted to other employees.

Company review systems were also lacking. Some 58 percent said there were no clear rules at their workplace on what must be checked before sharing AI-assisted work.

Use.AI stressed, "Along with the volume of content AI produces, the time and effort required to actually use the output in work must also be considered."

The survey was based on employees’ experiences, and it did not prove a net change in overall productivity from adopting AI or a causal link to increased rework.

As companies expand their use of AI, building work systems that verify the accuracy of output and clarify responsibility is emerging as a new task.