As the appeal of AI workforce transformation continues, a hidden threat is impeding expected productivity gains known as workslop. That is, quickly produced, low-quality work generated using AI that produces flawed results, adds minimal or negative value, and often requires extra human oversight and capacity to correct. In other words, the very tools meant to improve productivity can also drain it.
Workslop is often the result of employees being pressured to produce more work faster, leaving no time or autonomy to do proper quality checks. It’s also driven by misdirected effort and incentives that encourage frequency of use and increased output volume rather than higher-quality work.
Low-value AI output is making it harder for many organisations to realise value from AI investments, despite growing adoption across everyday workflows. It also undermines their ability to shape the workforce for an AI-enabled future.
Tackling workslop isn’t about stopping employee laziness or corner cutting, nor does it require technical solutions. Instead, it can be reversed by revisiting the strategy for how employees apply AI in their workflows and how they’re incentivised to behave. This involves identifying AI-friendly tasks and providing targeted use sessions, as well as following through with performance metrics that support future workflows.
Use AI where it adds value
With AI tools supporting specific tasks that were previously outsourced, many employees are not only producing more, they’re absorbing responsibilities that previously justified additional support.
While the notion of more work being done at a faster pace is certainly appealing, what’s perceived as enhanced productivity can lead to weakened decision making and cognitive fatigue, resulting in greater instances of workslop.
Many employees using AI experience an uptick in work friction, such as having to create new processes or workarounds for formal processes. What’s more concerning is that the frequency in this work friction increases the more frequently employees use AI. However, when AI is applied to tasks selectively, workflows can unlock adaptability, increase human bandwidth and improve decision quality.
Instances of workslop tend to increase in AI-friendly organisations not because people are cutting corners but because nobody considered the workflows and tasks best suited for AI.
Business leaders should carefully consider whether workflow changes aligned with using AI genuinely add value or merely introduce unnecessary complexity. Employees must also be consulted about AI workarounds and points of friction, which will determine where AI could potentially be helpful to their work or a hinderance.
Once tasks have been identified, it’s important to map out how each will change and what new skills are needed moving forward. This involves considering if capacity has changed, how jobs or workflows need to change to get the most out of AI investments and whether existing investments can be leveraged to automate tasks.
Targeted AI task development sessions should be rolled out to support workflow changes, which will teach employees how to best use AI in the context of their daily work. These sessions must be tied directly to the workflow changes employees helped design, as well as prioritised by how critical the skills are and the timeline for job transformation. These sessions should be measured through performance outcomes tied to new responsibilities.
Measure outcomes, not activity
As a result of heavy AI implementation across organisations, AI engagement-based indicators are often incorporated within individual performance metrics. This can include measures such as how often an employee is using an AI tool, as well as time-to-production goal.
While it’s important to examine how AI impacts the speed of task completion, it doesn’t mean speed should be incentivised in performance metrics at an individual level. More often than not, these engagement-based indicators only capture surface-level activity, failing to account for things like reduced errors, accelerated decisions or process improvements.
As a result, instances of workslop (or a reduction of it) remain obscured by metrics that speak to frequency, not quality of use. Defining how AI value is understood and executed can be achieved by updating performance metrics standards across the organisation, reducing the emphasis on speed at the cost of value creation.
Avoid any mandated AI usage performance metrics, by aligning them with organisational strategy and maturity using non-evaluative goals. Support leadership by creating guidelines for usage norms and set base and stretch targets once there is evidence of repeated improvements to facilitate continued learning.
Moving forward, the businesses most likely to achieve value from their AI investments will be those that help employees use AI tools to reduce friction, rather than simply cut down on time spent or increase output on daily activities.