Trust Levels Shape AI Adoption And Performance In Companies

A study published today in the Journal of Management Studies, offers valuable insights into how organizational members’ trust in AI shapes adoption and performance.

The research, based on a qualitative, real-life study, identifies four distinct trust configurations among employees: full trust, full distrust, uncomfortable trust, and blind trust.

These trust levels—comprising cognitive (belief-based) and emotional (feeling-based) components—affect behaviors in digital environments which, the researchers argue, will influence AI’s effectiveness within organizations.

Full trust is when employees have both high emotional and cognitive trust in AI. Uncomfortable trust refers to high cognitive but low emotional trust, while blind trust means employees have low cognitive but high emotional trust.

Different levels of trust have significant effects on how employees interact with AI in the workplace. Full trust in AI leads to smoother adoption, as employees feel confident relying on the technology for decision-making.

However, when employees completely distrust AI, they may resist its use, limiting its effectiveness and the potential benefits. Uncomfortable trust, where employees are unsure but still use AI, can lead to inconsistent results, as employees may not fully commit to the technology.

These behaviors created a “vicious cycle,” where inaccurate or incomplete data input lowered AI performance, further damaging trust in the technology and delaying its adoption.

On the other hand, blind trust—where employees trust AI without question—can cause issues, as it may overlook flaws in the system, leading to biased data or poor decisions.

The study suggests that trust is not just about how much employees cognitively understand AI, but also how emotionally comfortable they feel with it. Interestingly, the introduction of AI can disrupt trust levels even in organizations with a culture of transparency.

Employees’ emotional reactions to AI, such as fear or excitement, play a huge role in determining whether they’ll adopt it. The researchers developed a model that explains how different trust configurations lead to different behaviors, which in turn influence AI performance and, ultimately, AI adoption.

The model suggests that when employees feel trusted and comfortable, they provide more accurate data to the AI, which improves its performance. On the other hand, distrust leads to biased or limited data, damaging the AI’s ability to function effectively.

The study also calls for a more human-centered approach to AI adoption. Instead of focusing only on AI’s technical aspects, leaders should address both the cognitive and emotional concerns employees have about the technology. This means recognizing that different trust levels require different strategies to help employees adjust to AI.

Managers can improve AI adoption by building both cognitive and emotional trust. To build cognitive trust, it’s important to provide training to help employees understand how AI works, explain its capabilities and limitations, and set clear policies on data usage. Managing expectations and being patient with AI as it improves over time is also crucial.

To build emotional trust, managers should encourage open conversations about concerns, show excitement about AI’s potential, and create a safe space for employees to express their feelings. Leaders should also ensure AI is used ethically and responsibly, especially when handling sensitive data, to make employees feel secure.

In conclusion, the study emphasizes that trust plays a crucial role in how employees behave around AI. Understanding and addressing both the cognitive and emotional aspects of trust can help companies successfully integrate AI technology into their workflows.