A survey of 1,100 users of artificial intelligence (AI) tools in organizations with more than 250 employees in the U.S. finds that while nearly all (92%) have seen productivity improvements, they are also spending on average 4.5 hours per week cleaning up AI mistakes.
Conducted by the market research firm Centiment on behalf of Zapier, a provider of a platform for automating workflows and processes, the survey finds nearly three quarters (74%) have experienced at least one negative consequence from low-quality AI outputs, including work rejected by stakeholders (28%), security incidents (27%), and customer complaints (25%).
The survey makes it clear that while AI tools can be a boon for productivity, they are also increasing “workslop,” says Emily Mabie, senior AI automation engineer at Zapier.
Analytics tools specifically create the greatest amount of workslop, with 55% of respondents noting data analysis and visualizations require the most cleanup, followed by writing tasks (46%).
Overall, the survey suggests there is a need for additional training to reduce the amount of workslop being generated, says Mabie. Employees without AI training are 6x more likely to say AI makes them less productive (6% vs. 1%), the survey finds. While untrained workers spend less time on AI cleanup, they also report fewer productivity gains. Just 69% say AI helps, compared to 94% of trained workers.
The amount of AI workslop that needs to be cleaned up also varies widely by department. Engineering, IT, and data roles average 5 hours per week fixing AI outputs, with 78% reporting negative consequences. Finance and accounting teams face the highest rate of negative consequences at 85%, averaging 4.6 hours of cleanup per week. Workers spending more than five hours weekly on AI cleanup are more than twice as likely to report lost revenue, clients, or deals (21% vs. 9%).
In general, employees who have had AI training spend more time with AI and more time fixing it, yet they remain dramatically more likely to say it’s worth it, notes Mabie. They use AI more aggressively, more frequently, and in higher-stakes contexts where both the benefits and the cleanup requirements are greater, she adds. As such, there needs to be a lot more focus on AI fluency, notes Mabie. “AI training should be mandatory, not optional,” she says.
Additionally, employees with access to AI orchestration tools report the highest productivity gains at 97%. Those that can share additional context with AI tools, such as internal documentation, brand guides, and project templates, report achieving a 96% productivity boost. A full 95% of employees with access to prompt libraries and ongoing training say AI makes them more productive.
At this point, it’s unlikely AI will replace the need for employees any time soon, said Mabie. There is still a clear need for humans to be in the loop to successfully complete tasks, she adds.
There is little doubt that most employees are making use of AI to one degree or another. The challenge and the opportunity now is to drive the adoption of AI deeper into the organizations while minimizing the number of negative experiences that today are still all too frequent.
