TLDR: This research paper explores how air traffic controllers (ATCOs) develop and manage trust in their tools and colleagues within the safety-critical environment of UK air traffic control. Using an ethnographic approach, it reveals that trust in technology is not binary but dynamic, constantly calibrated through experience and shared knowledge. The study suggests that for AI tools to be effectively adopted, ATCOs need to understand their “trust contours” and learn when to rely on them and when to apply workarounds, rather than requiring perfect trustworthiness from the AI.
The adoption of Artificial Intelligence (AI) in organizational settings often faces significant socio-technical challenges, particularly concerning how people trust the tools they use daily. A recent ethnographic study, detailed in the paper “Trustworthy AI: UK Air Traffic Control Revisited”, explores these challenges within the safety-critical domain of air traffic control.
The study, which is part of Project Bluebird, aims to understand the role of trust in air traffic control work, including the trust air traffic controllers (ATCOs) have in each other, in their current tools, and in the broader air traffic control system. This understanding is crucial for the successful integration of new AI technologies, such as agent-based systems, which are being developed to manage aircraft safely.
Trust in technology is not a simple yes-or-no concept; rather, it is nuanced and constantly adjusted based on users’ lived experiences. This research introduces the idea of ‘boundaries,’ ‘contours,’ or ‘gradients’ of trust, highlighting that trust is dynamic and calibrated over time. In safety-critical fields like air traffic control, dependable system performance is achieved not just through reliable technology, but also through the practical actions and workarounds ATCOs employ when systems are less than perfect.
The methodology for this study involved ethnographic observations of ATCOs in UK air traffic operations rooms and during training sessions in simulated environments. Discussions were also held with ATCOs and instructors. The goal was to understand what it means to be a competent ATCO, how their work practices are shaped, what tools they use, and the role trust plays in their use of these tools.
One key finding is the importance of collaboration among ATCOs, who are organized into ‘watches.’ Members of a watch are co-located, facilitating formal and informal collaboration, such as sharing tips and experiences about using particular tools. Interpersonal trust among ATCOs is vital for the safe and efficient management of aircraft.
ATCOs’ primary objective is aircraft safety, ensuring minimum separation standards are maintained. While tools are available to assist them, including those that predict potential conflicts, ATCOs learn from experience when to trust these tools and when to exercise caution. They understand the “weaknesses” of their tools and can distinguish between dependable and undependable behaviors, often developing workarounds. For instance, an ATCO might learn to filter out “spurious conflict alerts” from a tool to identify real threats.
The study concludes that AI tools do not need to be perfectly trustworthy to be beneficial in air traffic control. What is essential is that ATCOs can recognize and navigate the “trust contours” and “boundaries” of these tools, knowing when and how to react at any given moment. This involves understanding the strengths and weaknesses of the AI, which can be fostered through training and transparent AI behaviors (explainable AI).
Also Read:
- Building Rapid Trust in AI: How Adaptive Explanations Can Transform High-Stakes Human-AI Teams
- Transparent AI Collaboration: SynLang and the Philosophy of Shared Intelligence
Ultimately, the research emphasizes that ATCOs continuously refine their skills in assessing tool trustworthiness through individual experience and shared knowledge within their watches. These informal processes are as crucial as formal training in building a robust “trust architecture” within the socio-technical system of air traffic control, ensuring safety and dependability even when facing everyday challenges.
