When AI Helps Us Learn and When Pressure Turns It into a Shortcut
Contributors: Cristina Maria Ionescu, Alexandra Kogalniceanu, Daniela Cismaru, Otilia Saracutu, Mihai Rotaru, Dr. Laura Baragan, Dr. Alexandru Ionescu, Cristina Cristescu, Adriana Tuluca-Marina
Organizations are investing heavily in artificial intelligence to improve productivity, accelerate work, and support learning. Yet one of the most important questions is not what AI can do, but how people actually use it in their daily work.
To explore this question, the Institute for Sustainable Human Performance organized a Sense-Making Lab on AI and Sustainable Performance, bringing together managers, consultants, HR professionals, educators, researchers, and practitioners from different sectors. Rather than focusing on technology itself, participants reflected on the organizational conditions that shape how AI is used and its effects on learning, performance, and human capability. The resulting insights are summarized in the accompanying Insight Brief.
The Same Technology, Different Outcomes
One of the strongest points of consensus was that AI use changes under pressure.
When workload increases, deadlines become tighter, and performance metrics emphasize speed, people tend to use AI primarily to execute tasks faster, summarize information, and reduce effort. Participants recognized clear productivity benefits, but they also identified an emerging risk: verification and reflection often decrease when speed becomes the dominant objective.
Interestingly, the same technology was perceived very differently when employees faced uncertainty rather than pressure.
When participants lacked expertise, explored unfamiliar topics, or sought alternative perspectives, AI became a learning partner. They used it to brainstorm, challenge assumptions, generate ideas, and structure their thinking. Under these conditions, AI was viewed as supporting capability development rather than merely accelerating output.
However, an emerging concern is overreliance on AI-generated outputs, particularly when organizations reward speed and efficiency above all else. This aspect was suggested by recent work by Leonardi and Leavell (2026), who describe the problem of artificial certainty. Their research shows that when AI-generated representations are presented as objective reality rather than as one possible interpretation, people become more likely to mistake representations for reality. As a result, uncertainty disappears from the conversation, and expert judgment becomes less visible. Importantly, Leonardi and Leavell argue that expertise increasingly involves helping others preserve uncertainty rather than eliminate it.
Participants in our Lab appeared to reach a similar conclusion. They emphasized the importance of maintaining human review, discussion, and accountability, particularly when decisions have significant consequences.
Learning Requires More Than Access to AI
Another important insight concerned critical thinking. Many participants appreciated AI's ability to support idea generation, summarize complex information, and provide alternative perspectives. However, they also recognized that these benefits depend on how people interact with the technology.
Recent experimental research by Gerlich (2025) provides a useful explanation. The study found that unguided AI use often leads to cognitive offloading, in which individuals transfer cognitive responsibilities to the system without improving the quality of their reasoning. In contrast, structured prompting encouraged deeper engagement, stronger critical reasoning, and greater reflective involvement.
This finding helps explain a recurring theme in the Lab discussions. Participants were not concerned that AI would replace thinking. They were concerned that organizational conditions might lead to resource depletion and ultimately to burnout. The challenge is therefore not simply teaching employees how to use AI, but helping them use it in ways that preserve resources.
Why Sharing Practices Matters
Participants strongly supported creating spaces where employees can share experiences, use cases, and lessons learned from working with AI.
This insight aligns with recent evidence suggesting that AI functions as a double-edged aspect of work design. Pinho, Fontes, and Santos (2025) found that collaboration with AI can increase engagement and work-life balance when supported by appropriate resources and learning opportunities. Their findings suggest that awareness of AI's challenges does not necessarily undermine engagement. Instead, under supportive conditions, awareness can become a catalyst for learning and adaptation.
Sharing practices may therefore play an important role in helping organizations transform AI from a source of uncertainty into a source of collective learning.
Institute Reflection: The Question Behind the Technology
One of the strongest messages emerging from the Lab was that organizations often focus on implementing AI tools before understanding the conditions shaping their use.
Yet participants consistently pointed toward a different challenge: pressure, workload, uncertainty, expertise, and performance expectations shape behavior long before technology enters the picture.
At the Institute for Sustainable Human Performance, we increasingly view AI adoption as an organizational design challenge rather than a technological one.
The question is not only how to introduce AI. The question is how to create environments where people continue to think critically, learn continuously, and sustain performance over time.
References
Gerlich, M. (2025). From offloading to engagement: An experimental study on structured prompting and critical reasoning with generative AI. Data, 10(11), 172.
Leonardi, P. M., & Leavell, V. A. (2026). Knowing enough to be dangerous: The problem of artificial certainty for expert authority when using AI for decision-making and planning. Organization Science.
Pinho, J. C., Fontes, A., & Santos, G. G. (2025). Balancing the double-edged sword of artificial intelligence: Job demands, resources, and work-life balance. Computers in Human Behavior Reports, 100924.
Comments