What is workslop? Stanford's finding that should change how you upskill for AI

What workslop actually is

Most employees have experienced it. The AI-generated report that covers all the headings and says nothing. The analysis with correct formatting and flawed logic. The summary that sounds thorough and misses the point.

Stanford and BetterUp Labs named the pattern: workslop. AI-generated output that masquerades as good work but lacks the substance to meaningfully advance the task.

The numbers behind it:

• 40% of employees report having received AI-generated workslop recently
• Each incident takes an average of two hours for another employee to correct
• The downstream effects extend beyond the immediate task - eroded trust, lower collaboration quality, weaker inputs reaching leadership

The hidden cost is not in the AI output itself. It is in what happens next. When someone produces workslop, they are not outsourcing cognitive work to a machine. They are passing it to a colleague, who now has to decode what is missing, correct what is wrong, and handle the situation without damaging the relationship.

Why upskilling programs aren't solving it

Most organizations have AI training in place. Most employees are still producing workslop.

The reason: 85% of workers say they cannot connect their AI skills training to their actual work.

The common pattern is programs that focus on tool access and prompt engineering, but skip the judgment layer - knowing what good output looks like, when to use AI and when not to, and how to interrogate a result before passing it on.

When leadership tells employees to use AI broadly without guidance on standards and appropriate use, people apply the tools without the discernment needed to make them useful. The tools get used. The work quality degrades.

The four stages where HR and L&D should be working

Addressing workslop starts with an honest view of where your organization is in its AI capability journey. The organizations seeing durable results move through four stages - and the work for HR and L&D looks different at each one.

‍

Stage 1: AI understanding 

Before employees can use AI well, they need to understand what it can and cannot do. This means assessing team readiness and building data and AI literacy alongside platform training, not after it. Skipping this stage is where most workslop begins.

‍

Stage 2: Individual productivity 

Once the foundation exists, people can start generating real gains from AI in their own work. L&D's role here is creating space for structured experimentation and measuring what changes in how people work - not just whether they completed a module.

‍

Stage 3: Work redesign

Individual efficiency plateaus until workflows change around it. This is where HR and L&D work with business leaders to identify which processes need redesigning. The gains from Stage 2 don't compound without this step.

‍

Stage 4: Organisational transformation 

Capability becomes competitive advantage when it spreads across functions - not when it stays in pilot teams. HR and L&D can build the case for cross-functional rollout using earlier stages as proof of concept, and connect the capability roadmap to business priorities rather than running it alongside them.

‍

Most organizations are stuck between Stage 1 and Stage 2. They have deployed tools and run training. They have not yet built the judgment layer that decides what to do with the output.

What this means practically

Three shifts that address workslop specifically:

• Teach standards, not just prompts. What does good AI output look like for this task? What signals should tell you it needs more work? These are teachable, but they require a different training design than most programs currently use.
• Create space for structured experimentation. Employees need practice interrogating AI outputs, identifying flaws, and building the reflex to check before passing work on. That practice rarely happens in a module.
• Measure output quality, not just tool usage. If the metric is "how many people are using Copilot," you will get high usage and unpredictable quality. If the metric is "what has changed in the quality of work AI supports," you get a different conversation.

The organizations closing the gap between AI adoption and AI quality are not necessarily the ones with the most training hours. They are the ones that treat discernment as a skill - coachable, measurable, and worth investing in.