From intuition to instruction: capturing the unspoken expertise of skilled workers [Forum special series, KM in construction]

This article is part of a special series of summaries of keynotes and presentations from the KM Triversary Forum 2025, and also part of the KM in the building & construction industries series.

This article is prepared as an accessible summary of my KM Triversary Forum presentation, “From Intuition to Instruction: Capturing the Unspoken Expertise of Skilled Workers.” It is based on my paper¹, “Unveiling the Untapped: Extracting Tacit Knowledge from Experienced Construction Workers through Visual Attention Interfaces,” presented in the Proceedings of the 23^rd CIB World Building Congress.

Why this matters

Every workplace has people who can sense trouble before others can explain what is wrong. On a construction site, an experienced worker may glance at a scaffold connection and immediately feel that something deserves attention. A novice may look at the same scene and see only a busy work area.

This is not magic. It is experience. Knowledge management calls it tacit knowledge: the hard-to-explain know-how people build through years of practice. It includes judgement, timing, pattern recognition, risk awareness, and the ability to read a situation quickly. It is often the knowledge behind statements such as: ‘I just know where to look’, ‘that does not feel right’, or ‘you learn it with experience’.

The challenge is that this knowledge is extremely valuable, but difficult to teach. Manuals, procedures, and checklists are essential, but they do not fully show how experts actually pay attention in real situations. My presentation explored a simple but powerful question: can we help novices learn to see what experts see?

The central idea: make attention visible

The study uses two technologies to turn part of expert intuition into something visible and discussable. Eye-tracking shows where a person looks. It can record fixations, which are moments when the eyes pause on something, and saccades, which are quick jumps from one point to another. When this information is overlaid on an image, it can produce a heatmap showing which parts of a scene attract the most attention.

EEG, or electroencephalography, records patterns of brain activity through sensors placed on the head. In this study, EEG helped indicate attention stability and peripheral attention – the ability to stay aware of things at the edge of vision, even when the eyes are focused somewhere else.

Together, eye-tracking and EEG do not ‘read minds’. They do something more practical: they provide observable clues about how expert attention works. They help us see the difference between simply looking at a site and professionally noticing what matters.

What the study did

The research compared experienced and novice construction workers in realistic visual tasks. There were 20 participants: 10 experienced workers and 10 novices. The experienced workers included supervisors, inspectors, and tradespeople with more than 10 years of experience. Novices had less than two years of experience.

Participants completed three construction-related tasks:

• Structural inspection – assessing scaffold setups and framing structures for stability, safety, bolts, supports, and connections.
• Site safety evaluation – identifying safety concerns in high-resolution construction site images within a limited time.
• Spatial planning and material placement – judging how materials should be placed to maintain clear routes, reduce clutter, and support efficient workflow.

The aim was not simply to ask experts what they knew. It was to observe how their attention behaves while expertise is being used.

What we found

Finding 1: Experts focus on what matters first

Experienced workers did not look everywhere equally. They spent more time looking at high-risk or high-value areas, such as structural joints, support beams, scaffold connections, and other critical components. Novices tended to scan more broadly and sometimes spent more attention on non-critical areas.

In simple terms, experts had better visual priorities. They knew where the important questions were likely to be hidden. This matters because training often tells people what to check, but not always how to build the expert habit of looking at the most important areas early.

Finding 2: Experts use both focus and peripheral awareness

One of the most interesting findings was that experts did not only stare at the obvious hazard. They also seemed to maintain awareness of what was happening around the main point of focus. This is the ‘centre + sweep’ pattern: inspect a critical detail but keep scanning the wider environment.

This is especially important in construction, where risk can emerge from movement, clutter, machinery, other workers, unstable materials, or changing site conditions. A safe worker not only looks carefully, but they also look carefully while staying aware of the surroundings.

Finding 3: Experts change their strategy depending on the task

Expert attention was not one fixed habit. It shifted depending on the work being done. In structural inspection, experts focused on load-bearing areas and connections. In safety evaluation, they combined direct checks with fast peripheral sweeps. In layout and materials tasks, they used more planned scan paths to judge access, movement, and placement.

This matters because expertise is not just more knowledge. It is flexible knowledge. Experts adapt how they look, think, and decide according to the situation.

A simple way to understand the difference

From intuition to instruction

The purpose of this work is not to reduce skilled workers to data points. It is to respect their expertise by finding better ways to preserve and transfer it.

A skilled worker may not be able to fully explain every cue they use, especially when judgement is fast and intuitive. But if we can capture where experts look, how long they attend, when they scan the periphery, and how their strategy changes across tasks, we can create practical learning tools.

For example, expert heatmaps can become visual playbooks. Novices can compare their scan patterns with expert patterns. Trainers can use these differences to ask better coaching questions: What did you look at first? What did you miss? Why did the expert check that connection? What did the expert keep monitoring while focusing on the main task?

This approach does not capture all tacit knowledge. Some expertise will always remain embodied, contextual, and social. But it can capture useful traces of expertise – enough to make invisible judgement more visible, discussable, and teachable.

Potential applications beyond construction

Although the study is grounded in construction, the idea has wider relevance. Many fields depend on experienced people noticing important signals quickly, often before they can fully explain their reasoning.

• Construction and infrastructure: Training apprentices to identify hazards, quality defects, sequencing issues, access constraints, and high-risk site conditions faster and more reliably.
• Healthcare and diagnostics: Helping trainees in radiology, surgery, nursing, emergency medicine, and paramedicine learn where experienced clinicians focus attention when reading images, monitoring patients, or assessing rapidly changing situations.
• Aviation, rail, and transport: Supporting pilots, drivers, signallers, and control-room operators to develop expert scan patterns, situational awareness, and early risk detection.
• Manufacturing and quality control: Capturing how skilled inspectors detect defects, identify abnormal machine behaviour, and prioritise checks in complex production environments.
• Mining, energy, and utilities: Improving safety training for high-risk environments where workers must watch equipment, terrain, tools, weather, alarms, and colleagues at the same time.
• Emergency management and disaster recovery: Helping firefighters, incident controllers, recovery officers, and emergency teams transfer hard-won situational awareness from experienced responders to newer personnel.
• Maintenance and asset management: Documenting how expert technicians inspect plant, buildings, infrastructure, and equipment to spot early warning signs before failures occur.
• Sport, coaching, and performance: Showing athletes and coaches how experts read the field, anticipate movement, and decide where to focus under pressure.
• Education and vocational training: Creating more practical forms of training in which learners do not only hear instructions but also see expert attention patterns and practise them.
• AI and digital decision support: Using expert attention patterns to train computer vision systems, digital assistants, or augmented reality prompts that highlight where human expertise suggests attention is needed. This should support, not replace, human judgement.

What this could look like in practice

A workplace that wants to apply this approach could begin with a small, ethical pilot rather than a large technology rollout:

1. Choose one high-value task where expert judgement matters, such as scaffold inspection, pre-start safety checks, defect detection, emergency triage, or equipment maintenance.
2. Record how several experienced workers visually approach the task using non-invasive eye-tracking and, where appropriate, EEG.
3. Create heatmaps and attention summaries that show common expert patterns, while protecting individual privacy.
4. Turn the patterns into training materials, such as visual playbooks, short videos, simulations, or VR/AR exercises.
5. Ask novices to complete the same task and compare their attention patterns with expert examples.
6. Use the comparison for coaching, not blame. The aim is to improve learning, not monitor workers punitively.
7. Measure outcomes such as time-to-competence, near-miss reduction, fewer rework hours, quality improvement, and learner confidence.

Ethics: this must be about learning, not surveillance

Any system that captures attention data must be designed with care. Workers should know what is being recorded, why it is being recorded, how data will be used, and how privacy will be protected. Consent, anonymisation, secure storage, and clear governance are essential.

The strongest use of this approach is not to judge whether a worker is ‘good’ or ‘bad’. It is to support learning, mentoring, and safety. Attention data should open better conversations between experts and learners, not replace trust, experience, or professional judgement.

What success might look like

If this approach works, success should be measured in practical outcomes. Do new workers become competent faster? Do they notice critical risks earlier? Are there fewer near misses, fewer defects, and fewer rework hours? Do supervisors find it easier to explain expert judgement? Do experienced workers feel that their know-how is being respected and preserved?

The next step is field validation. Laboratory tasks are useful, but construction knowledge is shaped by real sites, real pressure, weather, noise, teamwork, tools, materials, and changing conditions. Future research should test portable, non-intrusive tools in live environments and across different trades and industries.

The future of knowledge transfer may not be only about writing more manuals. It may also be about helping people see, notice, and attend like experts.

For knowledge management, the lesson is clear. Much of what organisations need to preserve is not only in documents or databases. It is in skilled perception, embodied judgement, and the quiet expertise of people who have learned through practice.

From intuition to instruction means building a bridge between what experts do naturally and what novices need to learn deliberately. If we can make even part of that bridge visible, we can support safer workplaces, better training, and more respectful knowledge transfer across generations.

Biography:

Patrick S.W. Fong is a prominent researcher and educator specialising in knowledge management, innovation, organisational learning, and project-based knowledge dynamics, with over three decades of distinguished experience across academia and industry. He is currently an Associate Professor in the School of Engineering and Built Environment at Griffith University, Australia, where his work advances the role of knowledge, learning, and innovation in improving project delivery, construction management, circular economy transitions, and built environment performance.

Patrick is recognised among Stanford University’s World’s Top 2% Scientists for career-long impact and has also been named among the Top 50 most influential people in tacit knowledge management, reflecting his sustained contribution to understanding how experience-based, often unspoken expertise can be captured, shared, and transformed into organisational capability. His recent work continues to expand this agenda, including research on tacit knowledge capture, lessons learned systems, knowledge valorisation, and the translation of skilled-worker intuition into practical guidance for industry.

Before joining Griffith University, Patrick served as Associate Director of the Knowledge Management and Innovation Research Centre (KMIRC) at The Hong Kong Polytechnic University, Vice President of the Hong Kong Knowledge Management Society (HKKMS), and a judge for the Most Admired Knowledge Enterprise (MAKE) and Most Innovative Knowledge Enterprise (MIKE) Awards. He is co-editor of the influential book Management of Knowledge in Project Environments (Elsevier, 2005) and is widely recognised for pioneering the first Knowledge Management Performance Index (KMPI) for construction firms in Hong Kong.

Patrick has published influential studies on knowledge evolution, tacit knowledge transfer, storytelling, organisational learning, and project-based knowledge management. His recent publications and projects further demonstrate his commitment to turning research into impact, including work on lessons management in disaster recovery and project environments, and knowledge-to-action frameworks that help organisations move from insight to implementation. Through his teaching, research, industry engagement, and thought leadership, Patrick continues to shape how knowledge is understood, valued, and applied across different walks of life.

Presentation resources: PowerPoint slides.

Header image source: Created by Dr Patrick S.W. Fong using ChatGPT (GPT-5.5 Thinking) and Python Pillow, CC BY-NC-ND 4.0.

AI statement: The substantive research, findings, analysis, and conclusions in this article are the author’s own.

Reference: