Stakeholder Mapping helps you identify all relevant actors connected to the AI system, including users, decision-makers, and external parties. It supports understanding of who to engage, their interests, their level of influence, how much the system impacts them, and the extent to which they can impact the system.

• Project scope and objectives
• Team members familiar with the context
• Optional: list of contacts

• Visual map of stakeholder roles and relationships
• Clear identification of user types and influencers

Eliciting expertise, whether by watching operators perform their tasks (observation), asking them to go through the tasks step by step to explain them in more detail (walk-through / talk-through), and talking as they go through the tasks explaining what they are thinking about (verbal protocols – used for more cognitive tasks).

• A realistic task scenario or use case
• Access to real users working on a specific task (e.g.: air traffic controllers working in a tower, or pilots in the simulator)
• A facilitator and note-taker or recording setup

• Notes on user behaviour and interaction patterns
• Identification of pain points, confusion or inefficiencies in the real work carried out by users

Gathers in-depth qualitative insights from users by combining prepared questions with open-ended discussion. Useful for uncovering needs, pain points, and contextual factors that may influence the design of the AI system.

• An interview guide with core topics and open-ended questions
• 1 interviewer + 1 participant per session + 1 notetaker
• Quiet space or online meeting tool
• Consent procedures and note-taking / recording tools

• Rich qualitative data
• Identification of user needs, goals, and pain points

Breaks down complex tasks into sub-tasks to better understand user goals, processes, and interactions. It explicitly represents what people need to do to achieve a goal, including actions and cognitive steps.

• A well-defined user task or goal
• 1 analyst
• SMEs (Subject Matter Experts) or access to relevant documentation
• Basic understanding of the task

• Task hierarchy diagrams (graphical representation of goals, tasks, and sub-tasks)
• Identification of bottlenecks or redundancies in existing workflows
• Qualitative insights into user activities and potential areas for design intervention

Capture diverse opinions and experiences by guiding a group conversation on a focused topic. Useful for exploring needs, validating concepts, or identifying potential issues.

• Skilled facilitator and a separate note taker
• 5–10 carefully selected participants
• Discussion guide or scenarios
• Neutral, comfortable space (it’s better to hold these face to face rather than online)

• Grouped user requirements
• Refined ideas or scenarios

To describe ‘thought simulations’, narrative accounts of users achieving goals with a product or system. It helps designers and users visualise current activities or predict how new tools will be used, stimulating creative design and prototyping.

• User research insights or knowledge of prospective users
• Defined user roles or “Personas” (optional but recommended)
• Understanding of user motivations and potential interactions

• Vivid descriptions of user experiences
• Identification of specific requirements for a new solution
• Enhanced understanding of potential problem areas
• Concrete, yet revisable, interpretation of design solutions

To detail multi-person tasks by sequencing each operator's actions, demonstrating information transfer between crew members, and how the team functions with available equipment and automation. It's especially useful for critical operations.

• Prior data from HTA, Critical Incident Technique, Observation, or Simulations.
• Discussion with operational personnel is essential for realism.
• Understanding of the task's timeline and potential goal evolution.

• Visualisation of task rhythm and required coordination.
• Identification of information discrepancies between operators and system state.
• Pinpointing key decision points, critical actions, and communication vulnerabilities.
• Highlighting areas of time pressure, stress, or significant uncertainty.

Engage end-users and stakeholders in shaping early ideas about AI roles, human–AI interaction, and system behaviours before detailed design begins.

• A clear design challenge or operational focus
• 1 facilitator + 1 note-taker or designer
• Stakeholders and/or representative end-users
• Materials for collaborative ideation (e.g., templates, sketching tools, cards)

• Shared understanding of operational needs
• Early concepts of the AI assistant’s purpose and interaction model
• Human-centred design foundations for the system

Low-fidelity prototypes, also called mock-ups or paper prototypes, are quick, simple representations of a design concept. They help designers gather early feedback on structure, content, and basic functionality without investing significant time or resources.

• Basic materials (paper, cardboard, sticky notes, markers, scissors) if going for a physical prototype. Prototyping software (such as Figma) if going for a digital one.
• A simple design concept or flow to prototype
• Users or stakeholders to test the prototype

• Early insights into usability and user expectations
• Feedback on layout, flow, and functionality
• Ideas for refinement before committing to detailed design or development
• For AI concepts, they help clarify how a feature might be explained or controlled by the user before technical implementation begins.

To design operational explainability (OpXAI) interfaces for Human-AI Teaming, especially in safety-critical contexts like aviation. It structures explanations at different levels of abstraction based on psychological distance, allowing users to progressively query information.

• An AI system requiring explainability.
• Understanding of user needs and operational context.
• Consideration of time available for information assimilation.

• Explanation interfaces tailored to specific operational needs to be added to the HMI mock-up.
• Information presented at appropriate levels of detail (e.g., from overview to in-depth data).

The Wizard of Oz method lets you test an interface that seems autonomous but is actually operated (in part or fully) by a human “wizard.” This is especially useful for designs involving AI or complex logic that is not yet implemented.

• A prototype (digital or physical) that users can interact with
• A facilitator to guide the session
• A “wizard” to simulate the system’s behavior from behind the scenes

• Early insights on how people interact with systems that rely on AI, automation, or real-time responses
• Validation of desired behaviors before building expensive back-end functionality
• A clearer picture of edge cases and user expectations

To evaluate an AI concept or system's current status against typical Human Factors requirements across various areas. It aims to identify strengths and areas for improvement at different project stages.

• A collaborative session with relevant stakeholders (e.g., AI developers, subject matter experts).
• An expert facilitator familiar with HAIQU and its covered areas.
• An AI concept or system, even if it's just a consolidated idea (a prototype is not strictly required).

• A clear overview of how the system addresses Human Factors requirements
• Data visualisations (e.g., spider chart) in a dedicated dashboard.
• A list of actionable items to improve the concept or system.

To proactively identify hazards and mismatches in human–AI teaming by exploring “what-if” deviations from expected operational behaviour, especially under degraded or uncertain conditions.

• Experienced HAZOP facilitator with aviation and risk expertise
• Multidisciplinary team: HF experts, AI developers, system users (pilots, ATCOs), data scientists
• Participation of concept developers or end users depending on system TRL
• Dedicated secretary for documentation
• Guideword list

• Identification of potential human–AI interaction failures
• Traceable link between hazards and design evolution
• Actionable mitigations and design recommendations

To systematically gather and sort investigation data in safety-critical systems, particularly focusing on the interactions between the human element and other system components. It helps identify potential sources of performance failure due to dysfunctional interactions.

• A socio-technical system for analysis.
• Understanding of the system’s components and their interdependencies.
• (If available) Data from incident/accident investigations or safety analyses.

• Systematic analysis and identification of problems.
• Highlighted relationships between human elements and other factors.
• Structured factual information for incident reports.
• Insights into reducing issue probability and applying detection/correction means.

Validate early concepts or system prototypes by leveraging the domain knowledge of operational experts. Helps identify gaps, unrealistic assumptions, and potential use issues before detailed design.

• A high-fidelity prototype, mock-up, storyboard, or concept description
• 1–3 domain experts (e.g., pilots, air traffic controllers)
• A facilitator with HF or UX background
• Structured script or key scenario prompts

• Expert insights on feasibility and usability
• Early detection of system-level risks or design blind spots
• Verbal feedback, annotated artifacts, structured notes

To analyse accidents and prevent failures by viewing them as control problems within complex socio-technical systems. It models the interrelationships between people, organisational structures, engineering activities, and system components, identifying where adaptive feedback functions fail to maintain safety.

• A complex, tightly coupled socio-technical system for analysis of designs.
• A high level of expertise in system theory and control systems.
• Resources for an in-depth analysis of regular processes and organisational structures.

• An overview of all links and interactions between system components.
• Identification of control errors that could lead to accidents.
• A model of the organisational safety structure.
• Insights for accident investigation, prevention, hazard analysis, and risk management.

Proactively identify and address potential legal issues, especially liability risks for providers, organisations and operators, that may arise from the use of AI-based solutions in aviation. It helps ensure that new technologies are not only safe and efficient but also legally sound and acceptable to all stakeholders.

• Detailed information about the operational concept and its context.
• Understanding of the AI system's level of automation and its impact on roles, tasks, and responsibilities.
• Relevant legal and regulatory frameworks (e.g., aviation law, EU AI Act, Product Liability Directive).
• SHS assessments (e.g., OSD, HAZOP analyses) for the system.

• Identification of potential liability risks for various stakeholders (human operators, organizations, technology providers).
• Analysis of how AI impacts existing roles and responsibilities, and the emergence of new ones.
• Actionable recommendations for design improvements and mitigation strategies to reduce legal exposure.

To evaluate an AI concept or system's current status against typical Human Factors requirements across various areas. It aims to identify strengths and areas for improvement at different project stages.

• A collaborative session with relevant stakeholders (e.g., AI developers, subject matter experts).
• An expert facilitator familiar with HAIQU and its covered areas.
• An AI concept or system, even if it's just a consolidated idea (a prototype is not strictly required).

• A clear overview of how the system addresses Human Factors requirements
• Data visualisations (e.g., spider chart) in a dedicated dashboard.
• A list of actionable items to improve the concept or system.

Simulate real-world conditions with human operators to design, test and refine new concepts, procedures, and system functionalities. Supports safe, repeatable evaluation of performance and user acceptance before implementation.

• Access to realistic simulation facilities (e.g., ATC or cockpit simulators)
• Licensed operators (e.g., controllers, pilots)
• Defined reference and future scenarios
• Human performance objectives and measurement criteria
• Observation plan and questionnaires (for post-experiment analysis)

• Objective human performance data (e.g., workload, task success, communication metrics)
• Subjective feedback on usability, trust, and acceptance
• Evidence to support design decisions and refinements

Understand how users gather and process visual information during tasks, uncovering insights that might not be verbalised. Eye tracking supports skills capture, task performance measurement, and workload assessment.

• Eye tracking equipment (head-mounted or screen-mounted)
• Specialist software for recording and analysing data
• Trained analyst to set up, calibrate, and interpret results
• Participants performing representative tasks

• Scan-path visualisations
• Insights into usability, information layout, and cognitive load
• Evidence for interface redesign or training needs

Measure operators’ mental and emotional states in real time, going beyond self-reports to capture unconscious reactions. NEURO-ID combines neurophysiological data with behavioural and incident data to map the Human Performance Envelope (HPE).

• Wearable neurophysiological sensors (EEG, ECG, EOG, GSR, etc.)
• Data acquisition hardware and software
• Data-mining and machine-learning tools
• Domain knowledge to interpret results in context

• Objective measures of mental workload, stress, vigilance, and engagement
• Human Performance Envelope (HPE) definition for each operator
• Insights for adaptive automation or real-time monitoring
• Links to incident and accident investigations

Eliciting expertise, whether by watching operators perform their tasks (observation), asking them to go through the tasks step by step to explain them in more detail (walk-through / talk-through), and talking as they go through the tasks explaining what they are thinking about (verbal protocols – used for more cognitive tasks).

• A realistic task scenario or use case
• Access to real users working on a specific task (e.g.: air traffic controllers working in a tower, or pilots in the simulator)
• A facilitator and note-taker or recording setup

• Notes on user behaviour and interaction patterns
• Identification of pain points, confusion or inefficiencies in the real work carried out by users

Leverage system-generated data to understand real user behaviour and system interaction patterns.

• Set up and access to structured logs (time-stamped user actions, system events)
• HF analyst or data-savvy team member
• Analysis tool (e.g., spreadsheet, Python, dashboard)

• A timeline of user–system interaction
• Visibility into AI model outputs and decision points
• Evidence for system refinement or HF evaluation

Explore users’ thoughts, experiences, and reactions following a prototype interaction or validation session. Helps uncover usability issues, cognitive workload, and unmet expectations.

• A completed system interaction or validation session
• Interview protocol with open-ended prompts
• 1 facilitator (with basic interview skills)
• Recording or note-taking setup

• Rich qualitative data (quotes, themes, behavioural insights)
• Understanding of what worked, what didn’t, and why
• Input to guide refinements to the prototype or system

Capture different angles of the user experience in a structured and measurable way following a system interaction, to support validation and improvement.

• A completed validation session or exercise
• Appropriate questionnaire(s) selected
• Printed or digital forms
• Basic understanding of each scale’s focus

• Quantitative and qualitative feedback on system performance and user experience
• Comparative data across sessions or user types
• Inputs for refining usability, workload, and acceptance

Identifying and monitoring perceptions and attitudes toward use of AI-based systems through a 23-question questionnaire. Use it from early stages of design throughout the whole development cycle, involving a diverse range of stakeholders, from concept owners to end-users. It allows proactively to spot relevant aspects that could enable or inhibit the success of the proposed solutions.

• The Societal Acceptance Questionnaire in a document (either digital or printed)
• A diverse group of relevant stakeholders who have been introduced to the system
• An interviewer is optional but recommended to guide the participant and ensure the process is effective

• Quantitative and quantitative data on eight key dimensions of societal acceptance
• Identification of relevant societal aspects that could enable or inhibit the success of the proposed solutions

Takes a snapshot of safety culture perceptions of aviation personnel who have experienced using an AI-based digital assistant, e.g. in a simulator. This is prospective safety culture, but can still yield useful insights into how pilots/ATCOs feel about future HAT concepts, and how it could affect their safety culture, whether positively or negatively. It can also give an overall perception of the desirability of the IA from the user's perspective

• The safety culture question set
• 1 interviewer
• 1 participant per session
• questions on an iPad using a simple survey platform such as SurveyMonkey
• Quiet space or online meeting tool
• Consent procedures and note-taking/ recording tools

• Rich qualitative data
• Identification of user’s opinions and perceptions about the impact of the HA concept on future safety culture

Visualise how users actually experience the AI assistant in real settings: what they do, think, and feel, to identify mismatches, breakdowns, and new needs.

• Real-world data from interviews, observations, or analytics
• A defined scenario and user goal
• 1 facilitator + cross-functional team
• Journey map template (if applicable, to make things easier)

• A map of real user behaviour and sentiment across phases
• Identified pain points and unmet expectations