AI made the work faster. Or did it?

Gartner just released their "9 Future of Work Trends for 2026" guide for CHROs. One trend caught my attention: "AI workslop becomes organizations' top productivity drain."

Apparently "workslop" is the new word we should be using to describe when the work we give to AI is riddled with errors and hallucinations. Work that looks done, but isn't actually usable without significant human cleanup.

Gartner's research found that each incident takes nearly two hours to detect, diagnose, and fix.

The issue is that 84% of HR leaders say their organizations are using GenAI tools. Meanwhile, 91% of IT leaders say their organizations dedicate little to no time monitoring what those tools actually do to work quality.

Nearly everyone has rolled out AI in one way, shape or form. Almost nobody is systematically checking if it's working as expected.

Their recommendation? "Survey employees to understand the sources and prevalence of workslop in your organization."

The Intelligence to Solve the Problem DOES Already Exist

The employees know.

They're the ones who spot the plausible-sounding summary that misses the actual point. They notice the customer email that's polite but doesn't answer the question. They spend two hours finding and fixing what looked done.

The intelligence exists. It's already distributed across your organization, sitting in the daily experience of the people doing the work.

So asking the workforce IS good advice. The challenge is how to do this effectively.

Traditional Survey Tools Aren't Up for the Task

They ask the wrong questions.

Most surveys default to quantitative scales:

"On a scale of 1-5, how effective are our AI tools?"

"On a scale of 1-5, how much time do AI tools save you?"

"On a scale of 1-5, how satisfied are you with AI output quality?"

The result is average scores and dashboards, but no clarity. You still may not know where the two-hour cleanup cycles are happening or why.

All the questions can't be asked at once.

The instinct is to build a comprehensive survey that covers every angle. Ten questions. Twenty questions. Cover all the bases.

But finding the right solution requires diagnosing the problem first. Question 1 needs to be asked, responses analyzed, and then Question 2 can be informed by what was learned.

If most people report that AI-generated customer emails are creating rework, the next question isn't about code quality or sales proposals. It's: "What specifically is wrong with the AI-generated customer emails?"

That's adaptive intelligence gathering. Traditional surveys can't do that. They're built to ask everything once and generate a report.

Qualitative responses are hard to analyze meaningfully.

When the right question gets asked and responses come back, what happens next?

The typical move is to copy them into ChatGPT or Claude to find themes. But generic AI doesn't know your business context, strategic priorities, or what happened when similar issues came up three months ago. The result is generic themes that someone still has to interpret and prioritize.

That's the workslop problem showing up in the solution.

Survey fatigue is real.

According to Gartner research, employee engagement survey response rates have declined significantly, with many organizations seeing participation drop below 65%. When employees don't trust that their feedback will lead to change, they stop participating.

And asking about AI effectiveness right after leadership announced AI as a productivity breakthrough creates a credibility problem. Employees wonder: are we really open to hearing this isn't working, or is this survey checking a box?

Insight without action destroys trust.

Even when the analysis gets done and the top five sources of workslop are identified, what happens next?

If those insights sit in a deck that gets reviewed once and forgotten, employees learn that sharing what they know is pointless. Survey platforms deliver data but don't turn intelligence into action.

So if traditional surveys can't solve this, what can?

What the Right System Would Need to Do

Gartner found that business units that redesign how work gets done with AI are twice as likely to exceed revenue goals compared to those that just optimize individual tasks.

But "redesigning how work gets done" requires knowing where the current design is breaking. And that requires a different kind of listening.

Not "how do you feel about AI?"

But "where is AI creating work that looks done but isn't?"

That's not a survey question. That's a targeted intelligence request.

To actually solve workslop, an organization would need a system that can:

Ask one focused question when it matters most.

When a quality metric dips or a manager mentions rework is increasing, deploy a single targeted question to the people who see the problem.

Then automatically analyze those responses in the context of business objectives and operational priorities to surface patterns:

"Sales proposals have wrong pricing."

"Code reviews taking longer."

"Customer emails missing context."

Leadership now knows exactly where to look.

Guarantee psychological safety

People will tell the truth about where AI is creating problems if they trust their input won't be traced back to them. Real anonymity isn't a nice-to-have. It's the difference between "everything's fine" and "here are the three places AI output is creating downstream chaos."

Turn intelligence into action, fast

Insight without action kills trust. When someone reports "AI-generated reports look good but the numbers are often wrong," leadership needs to act. Adjust the workflow. Add a verification step. Train people on what to check.

Then tell people what changed because of their input. That's how organizations build systems where people keep sharing intelligence instead of giving up.

What Wasn't Said Out Loud

The workslop is happening. Employees see it. They're catching the errors. They're doing the rework. But without a system to surface what they know, that intelligence stays trapped.

Meanwhile, quality degrades quietly. Customers notice issues before leadership does. Competitors move faster because they figured out where AI helps and where it hurts.

Without a systematic way to capture workforce intelligence on specific operational problems, organizations discover issues the same way they always have: after the damage is done.

If you're interested in learning more about how to systematically capture and act on the collective knowledge of the workforce, take a look at lyssin.com or email Josh at josh.frantz@lyssin.com