43% of Workers Have Submitted AI Output They Suspected Was Wrong
GoTo's 2026 report: 43% of workers submit suspect AI output. Two-thirds say reviewing AI 'workslop' adds to their workload. Here's what to do about it.
Everyone measured the time saved. Nobody measured the time created.
Everyone tracked how many hours AI saves. Almost nobody tracked the hours it costs. GoTo's Pulse of Work 2026 report — 2,500 employees and IT leaders surveyed across ten countries — puts a number on the other side of the ledger: 66 per cent of workers say reviewing colleagues' AI-generated output adds to their workload. Seventy-seven per cent say AI-produced work actually takes longer to review than human-produced work.
The report has a name for this phenomenon: workslop. And it's quietly eroding the productivity gains that justified the AI investment in the first place.
The quality gap hiding behind the speed
The finding that should stop every business owner mid-scroll: 43 per cent of workers have knowingly submitted AI-generated content they suspected contained errors or fabricated information. Not because they're careless. Because they're under pressure — 60 per cent of employees report feeling pushed to use AI for productivity gains, and the tools make it easy to generate plausible-looking output at speed. The creating is fast. The checking is slower than doing the work manually.
This isn't fringe behaviour. Seventy per cent of employees now use AI for high-stakes tasks — 41 per cent for legal and compliance work, 31 per cent for safety-impacting decisions, 29 per cent for strategic planning. Nearly one in four IT leaders say AI-related mistakes have already affected customers, clients, or the bottom line.
43%
Submit AI output they suspect is wrong
GoTo Pulse of Work, May 2026
66%
Say reviewing AI work adds to their load
77% say it takes longer than human work
1 in 4
IT leaders report AI errors hit customers
Or affected the bottom line
What this looks like in Australian businesses
The survey is global, but the pattern maps directly onto Australian workplaces. We've written before about the AI literacy gap here — 84 per cent of Australian workers use AI, according to RMIT and Deloitte, but only 7 per cent use it well. The GoTo data shows what happens when that gap meets real workflows.
For an accounting practice: a junior uses AI to draft a tax research memo or reconcile transactions. The output reads well — polished, properly formatted. The partner reviewing it faces a choice. Trust the AI-assisted work and risk the error rate, or scrutinise it more carefully than a human-drafted memo and spend more time, not less. Either way, the firm loses.
For a trades business: AI-generated quotes with plausible but incorrect material quantities. Scheduling that optimises for an algorithm's logic rather than local road conditions. Customer communications that sound professional but promise something the technician can't deliver. Each one reads well on screen. Each one creates rework when reality disagrees.
The governance gap — and the skills erosion behind it
Only 44 per cent of companies have a formal AI policy, according to the report. Among those that do, 77 per cent of employees and 47 per cent of IT leaders say the policy needs improvement. The tools arrived faster than the rules.
Then there's the longer-term risk. Thirty-nine per cent of workers say AI overreliance is actively eroding their skills. Thirty per cent say they cannot function without it. Among Gen Z workers entering the workforce now, 46 per cent believe AI dependency is making them less intelligent. This isn't just a quality issue. It's a business continuity issue. If your estimator can't sanity-check a quote without a model, if your analyst can't write a client memo from scratch, you've traded speed for fragility.
Three things to do this month
First, add a verification step to every AI-assisted workflow that touches clients, safety, or money. Not a bureaucratic review layer — a 60-second domain-expert check on the output before it leaves the building. The data shows AI work takes longer to review than human work, which means the value of AI is in the first draft, not the final product. Treat it that way.
Second, write a one-page AI policy. Which tasks AI can and cannot be used for, which outputs require human sign-off, what happens when someone spots an error. Fewer than half of companies have any policy at all. Yours should.
Third, preserve the ability to work without AI in every critical function. The 30 per cent of workers who say they can't function without it represent a single point of failure in your business. AI is an accelerator, not a replacement for the competence underneath.
Key takeaways
Sources
▶Assumptions & methodology
- GoTo's Pulse of Work 2026 report surveyed 2,500 respondents (1,250 knowledge workers and 1,250 IT decision-makers) between November 2025 and January 2026, in partnership with Workplace Intelligence. The sample covered ten countries: the US, Canada, UK, Ireland, Germany, Austria, Switzerland, India, Mexico, and Brazil. Australia was not included, though the patterns align with Australian data on AI literacy and adoption gaps from RMIT/Deloitte and National AI Centre reporting.
- The 43% figure refers specifically to workers who used AI-generated content 'even though they suspected it was low quality or might contain errors or fabricated information,' per the report's wording. This represents acknowledged rather than total error rates — actual rates of undetected AI errors may be higher.
- The 77% figure comparing AI review time to human review time reflects employee perception rather than measured time-on-task data. Actual review time differences likely vary by task complexity with higher-stakes work requiring proportionally more scrutiny.
- The 84% Australian AI usage and 7% advanced user figures referenced in the article are from the RMIT Online / Deloitte Access Economics 'Beyond Prompting' report (March 2026), covered in a previous Field Note.
Next
Only 4% of Australians Trust AI Companies With Their Data
Field Notes are general commentary on AI trends for Australian businesses. They don’t constitute professional advice. Talk to your accountant, lawyer, or IT adviser before acting on anything specific to your situation — or talk to us if you want help working out where AI fits.
Worried your team's AI use is creating more work than it saves?
A short conversation can map where AI is adding value in your business and where it's quietly generating rework. Book a call.
Book a call →