A new feeling about AI is spreading

A new feeling regarding AI is spreading

AI is currently being aggressively introduced in many places. People expect significant productivity gains from it. Yet, in some teams, exactly the opposite seems to be happening.

Team performance drops. Team members gradually lose confidence in their own judgment, while trust in one another fades at the same time. What is happening there is difficult to grasp at first.

The pattern is often similar: For example, a marketing team uses an AI tool to analyze the results of a campaign. The generative AI recommends reallocating the budget – naturally based on a comprehensive data analysis. But then the team realizes that the AI fundamentally misunderstood the customer segments. The campaign fails, the budget is gone. Suddenly, everyone on the team doubts not only future AI recommendations, but also their own ability to judge them correctly.

When AI suddenly creates more work

What is often underestimated during the introduction of AI: It is not just about technical challenges. Team dynamics also change. A new pattern that becomes visible in the process is so-called workslop.

Workslop refers to AI-generated work outputs that look professional and complete at first glance, but do not substantially advance a task. For example, they contain too little context, remain superficial, or are insufficiently vetted. The consequence: Colleagues have to contextualize, correct, or completely recreate the content. So work does not simply decrease – it is passed on to others. This costs time and can strain trust within the team.

How widespread this phenomenon is is shown by a survey by BetterUp Labs and the Stanford Social Media Lab among 1,150 workers in the US: 40 percent stated that they had received workslop in the previous month. Respondents estimated that processing such an output takes nearly two hours on average. The figures are an indication of the potential scale; they cannot be applied directly to German companies without further context.

When AI suddenly creates more work - To the article

Workslop: When something looks good that helps no one

This is not just about faulty AI outputs. Collaboration can also suffer: Those who regularly receive unfinished or inappropriate results may wonder whether their colleague worked carefully – or whether the AI was simply used unchecked. And anyone who is no longer sure when an AI recommendation is sound and when their own judgment is required may become more hesitant to question it.

This creates trust ambiguity: A team believes it can or should have trust in a situation, but does not truly feel it. The uncertainty is then not only directed at the AI. It also affects one's own judgment and collaboration with one another. Harvard Business Review authors Jayshree Seth and Amy C. Edmondson describe how AI-related mistakes amplify such doubts and can make it harder to address uncertainty openly within the team.

What can leaders do?

First: Clarify the purpose. Not every task becomes better when AI is used. Teams should jointly determine where AI genuinely provides support – and where human experience, contextual knowledge, or a personal conversation are more important.

Second: Define quality together. What must an output contain for it to be truly usable for the next person? This includes, for instance, a clear question, relevant background, verified statements, and a concrete next step. This makes it evident when an output is still a draft and when it can be handed off.

Third: Do not delegate responsibility to the AI. Whoever uses an AI output remains responsible for reviewing and contextualizing it. Especially with far-reaching recommendations, teams should ask: What assumptions is the output based on? What is missing? What experience or knowledge from the team must be factored into the decision?

Fourth: Be able to talk about mistakes. If an AI recommendation does not hold up, the team should not immediately look for someone to blame. The key is to clarify together what was overlooked and how the work process can be improved. This requires a framework in which employees can express doubts, question results, and also reflect on their own use of AI.

Leaders should therefore not only ask: "Does the AI work?" An equally important question is: "How does working with AI change our team – and what do we need to continue trusting each other and our own judgment?"

Because trust is essential for team performance. AI can alter this dynamic. That makes a conscious understanding of how human-AI teams collaborate all the more critical. AI does not become productive simply by generating outputs. It becomes productive when those outputs help the team make better decisions – and strengthen collaboration.