The technology that changes trust at work
When everyone has their own assistant, they stop asking the person next to them. That does not show up in any productivity metric, and it is what changes an office the most. The paper describes two futures and does not choose. I see one of them happening.
They identified three sociotechnical traits of generative AI (multiple usability, unpredictable plausibility and hyper-personalization) that change how employees connect, share trust and share expertise. In a laissez-faire scenario, those traits fragment resource flows, creating "polymaths", "oracles" and "sirens" that erode cohesion. In a cultivation scenario, AI is deliberately integrated and strengthens the social network.
Executives are watching generative AI tools show up on desks and in team chats overnight, and with them comes the question of whether these algorithms will strengthen collaboration or fray the web of trust that holds the organization together. The study published in *Strategic Organization* starts from that concern and, through a sociotechnical analysis of real cases of GenAI adoption, breaks the technology down into three key traits: multiple usability, unpredictable plausibility and hyper-personalization. The authors lay out two contrasting scenarios, a laissez-faire model, where use is left to each employee's discretion, and a cultivation model, where AI is deliberately inserted into workflows, to observe how those traits reshape everyday paths of interaction.
In the laissez-faire scenario, multiple usability lets anyone turn AI into their own go-to tool, which produces "polymaths" who hoard scattered knowledge, "oracles" who dictate decisions without validating where they come from, and "sirens" who draw attention with convincing but often wrong answers. That triangle of roles fragments resource flows: trust becomes selective, expertise concentrates in a few nodes and collegiality weakens, creating invisible silos that erode organizational cohesion.
In contrast, the cultivation scenario shows that when AI is integrated through clear protocols, training and interaction metrics, the same traits strengthen the social network. Hyper-personalization lets the tool adapt to each team's working style, while unpredictable plausibility becomes a source of unexpected ideas that are validated collectively. AI then acts as a catalyst that widens the circulation of expertise and reinforces mutual trust, keeping the resource structure fluid and resilient.
The difference between the two worlds is not a matter of technology but of governance. Leaders need to move from a passive stance to active design: define when and how AI steps into processes, set trust and expertise indicators that detect deviations, and create feedback spaces where users can calibrate the plausibility of outputs. Only with that deliberate approach will generative AI stop being a threat to cohesion and become a structuring element of the organizational social fabric.
Conclusions
- Generative AI reconfigures the paths of workplace interaction, changing how trust and knowledge are distributed.
- Without clear guidance, the tool concentrates information in a few users and weakens collaboration.
- A deliberate integration strategy lets AI strengthen organizational cohesion and improve collective productivity.
- Leaders must choose between a laissez-faire and a cultivation approach to control the social impact of AI.