Workers are spending over three hours a week cleaning up AI ‘workslop’, raising concerns and questions about the real effectiveness of AI use across different sectors and jobs.

This is according to new data from Zapier, which surveyed 1,100 enterprise AI users to find the reality of using AI at work.

While ‘AI slop’ is being used more commonly to describe the use of AI to create mostly nonsensical videos and images at a large scale for distribution across social media sites, ‘AI workslop’ carries more stakes – this is AI-produced work that may look fine at first glance, but often requires revision and reconfiguration, potentially requiring more time to correct than to just produce the result in the first place.

Zapier found that 58% of those using AI at work spent over three hours a week revising or redoing AI outputs.

Despite this, 92% still say that AI boosts their productivity, showing that people still see a clear pay off in using AI at work.

Still, 74% say they have experienced at least one negative consequence from low-quality AI outputs.

Investigating further, Zapier found that data analysis was the biggest source of AI workslop, not writing.

When looking more specifically at how workers are viewing AI, Zapier found that untrained workers were six times more likely to say that AI makes them less productive.

Providing workers with the right skills to make their AI use more effective is therefore paramount to their experience in using AI on the job.

Recommended

• Meta Announces Plans For World’s Largest Sub-Sea Cable
• 65% of Employees Are Excited to Use AI at Work
• Seven In Ten UK Employees Using Unapproved AI Tools, Microsoft Finds
• Report: AI Adoption Elevates Human-centric Skills
• Will AI Enable Workers to Secretly Juggle Multiple Jobs?

Further, how AI is integrated into their work, and what tools are offered, will determine how workers review their relationship with AI.

Almost all (97%) of those with access to AI orchestration tools say that AI boosts productivity. Those with prompt templates, libraries, and ongoing training report similar levels of productivity enhancement (95%).

Without training, guidelines, or libraries, and just about three quarters (77%) say that AI helps them, which is still a decent figurehead for AI integration.