Workforce AI Readiness Diagnostic
18 questions across 8 people dimensions, one at a time. About ten minutes. A score per dimension, a 90-day plan for your two weakest, and a literacy target for each role.
Listen to this briefing
The Workforce AI Readiness Diagnostic
The people side of AI adoption
Most AI readiness checks spend one or two questions on people and the rest on data and technology. In the programmes I have worked on, the people side is where adoption most often stalls. Staff hide their use, managers have no time to coach, and saved hours vanish into extra work. Nobody has told anyone what happens to their job.
This diagnostic looks only at that side. Answer for the organisation as it works today, not as the policy describes it.
Eight dimensions, each scored out of 100, averaged with equal weight
- 01Leadership sponsorship
- 02AI literacy by role
- 03Psychological safety and fear
- 04Job and task redesign
- 05Verification and automation bias
- 06Incentives and recognition
- 07Training access
- 08Change capacity
Each question has four answers, scored 0 to 3. A dimension scores its share of the maximum.
Dimension bands, 0 to 100
The overall score is the average of the eight, so a dimension with three questions counts the same as one with two. Overall tiers: Fragile 0 to 39, Uneven 40 to 59, Building 60 to 79, Ready 80 to 100.
No sign-up. Your answers stay in your browser, and you can leave and pick up where you stopped.
Leadership sponsorship
Question 1 of 3 in this dimension
Who sponsors AI adoption across the workforce?
Leadership sponsorship
Question 2 of 3 in this dimension
Do senior leaders use AI tools in their own work, where staff can see it?
Leadership sponsorship
Question 3 of 3 in this dimension
Has leadership said in writing what happens to roles and to the time AI frees up?
AI literacy by role
Question 1 of 2 in this dimension
How far has AI literacy reached across the roles in your organisation?
AI literacy by role
Question 2 of 2 in this dimension
Could a typical employee explain what the AI tools they use can and cannot reliably do?
Psychological safety and fear
Question 1 of 3 in this dimension
Do people say openly when they have used AI on a piece of work?
Psychological safety and fear
Question 2 of 3 in this dimension
If someone finds an AI error, including one they let through, can they report it without penalty?
Psychological safety and fear
Question 3 of 3 in this dimension
How do people talk about AI and their own jobs?
Job and task redesign
Question 1 of 2 in this dimension
Have you mapped which tasks within roles AI will change?
Job and task redesign
Question 2 of 2 in this dimension
When AI changes a task, are role descriptions, workloads and targets updated?
Verification and automation bias
Question 1 of 2 in this dimension
Do staff know which kinds of task their AI tools tend to get wrong, and check more carefully there?
Verification and automation bias
Question 2 of 2 in this dimension
Is there a defined check before AI-assisted work leaves the team or reaches a customer?
Incentives and recognition
Question 1 of 2 in this dimension
Do performance measures and recognition reward using AI well, including catching its mistakes?
Incentives and recognition
Question 2 of 2 in this dimension
What usually happens to time that AI saves?
Training access
Question 1 of 2 in this dimension
How much practical AI training has a typical employee received?
Training access
Question 2 of 2 in this dimension
Do staff have approved AI tools they can practise on with their real work?
Change capacity
Question 1 of 2 in this dimension
How many major change programmes are affecting the same staff right now?
Change capacity
Question 2 of 2 in this dimension
Do line managers have the time and support to coach their teams through AI changes?
Score by dimension
Each dimension is scored as a share of its maximum. Below 40 is a gap, 40 to 59 is weak, 60 to 79 is adequate and 80 or above is strong. The overall score is the average of the eight, each weighted equally.
Your next 90 days
The two lowest-scoring dimensions, with a plan for each. Work on these before widening any rollout.
Literacy target by role
The level each role needs, what it looks like and how to check it. Priority comes from your scores on the dimensions that most affect that role.
| Role | Target level | Able to | Check | Priority |
|---|---|---|---|---|
| Board and executive team | Informed sponsor | Decide where AI may and may not be used, read an AI risk report, and use the tools on their own work. | Each executive can describe one personal use and the organisation's main AI risks. | |
| Function heads | Redesign owner | Identify the tasks in their function that AI changes, set the checking standard, and adjust targets. | A task map and updated targets exist for each affected role in the function. | |
| Line managers and team leads | Coach | Coach their team on AI use, hold the conversation about jobs, and spot over-reliance on output. | Every team member has had a conversation about how AI affects their role. | |
| Office and knowledge staff | Practitioner | Use approved tools on routine tasks, know where the tools are weak, and check output to match the stakes. | They pass a short practical exercise that includes finding an error in AI output. | |
| Frontline and operational staff | Aware user | Understand which decisions about their work involve AI and know how to question or report an AI error. | They can name the AI tools that affect their work and the route for reporting a problem. | |
| HR, risk, legal and audit | Assurer | Test AI use against policy, law and contracts, handle AI incidents, and adjust performance frameworks. | A tested incident route and performance measures that reflect redesigned roles. | |
| Technology and data staff | Builder | Configure and build AI tools with evaluation, logging and human checkpoints designed in. | Each deployed tool has an evaluation record and a defined human checkpoint. |
Turn the plan into a programme
The report sets out your dimension scores, both 90-day plans, the role table and the method on printable A4 pages. This is a self-assessment. It is not an audit or an employee survey, and its results depend on how candidly it was answered. Your answers are saved only in this browser and are not sent to me or anyone else.
What the research says
These are the published findings behind several of the questions. Each is quoted from the linked source. None of them feeds into your score.
| Finding | Source and sample |
|---|---|
| 52% of people who use AI at work are reluctant to admit using it for their most important tasks, and 53% worry that using it on important work makes them look replaceable. | Microsoft and LinkedIn, 2024 Work Trend Index (May 2024). 31,000 knowledge workers in 31 markets. Source → |
| Only 39% of people who use AI at work have had AI training from their company. | Microsoft and LinkedIn, 2024 Work Trend Index (May 2024). 31,000 knowledge workers in 31 markets. Source → |
| The share of employees who feel positive about generative AI rises from 15% to 55% with strong leadership support. Only about a quarter of frontline employees say they receive that support. | BCG, AI at Work 2025: Momentum Builds, but Gaps Remain (June 2025). More than 10,600 leaders, managers and frontline employees in 11 countries and regions. Source → |
| Only one third of employees say they have been properly trained, and regular use is sharply higher among employees who receive at least five hours of training. | BCG, AI at Work 2025: Momentum Builds, but Gaps Remain (June 2025). More than 10,600 leaders, managers and frontline employees in 11 countries and regions. Source → |
| In Singapore, 45% of workers are uncomfortable admitting AI use to their managers, with fear of being seen as less competent the top reason, and 63% have spent less than five hours in total learning to use AI. | Slack Workforce Index, Singapore release (December 2024). 17,372 desk workers, 1,008 in Singapore, fielded August 2024. Source → |
| On a task chosen to sit outside the model's capability, consultants using GPT-4 were 19 percentage points less likely to produce a correct solution than those working without it. | Dell'Acqua et al., Navigating the Jagged Technological Frontier, Harvard Business School Working Paper 24-013 (September 2023). 758 BCG consultants in a pre-registered experiment. Source → |
How the diagnostic is built
Each rule in the scoring is traced to its source. The thresholds and role targets are my own calibration and are marked as such.
Show the rules and their sources
| Rule | Source |
|---|---|
| AI literacy is set by role, taking into account each person's knowledge, experience and the context the AI system is used in | EU AI Act (Regulation (EU) 2024/1689), Article 4, applicable from 2 February 2025 Source → |
| Leadership sponsorship is scored on visible use and a written commitment, as well as on a named sponsor | BCG, AI at Work 2025, on leadership support and employee sentiment Source → |
| Hidden AI use is treated as a sign of fear, and open disclosure as a sign of safety | Microsoft and LinkedIn, 2024 Work Trend Index; Slack Workforce Index, Singapore release, December 2024 Source → |
| Five hours of practical training is the threshold in the training question | BCG, AI at Work 2025, on training hours and regular use Source → |
| Checking is scaled to the stakes and to how familiar the task is, because accuracy falls on tasks outside the model's capability | Dell'Acqua et al., HBS Working Paper 24-013, 2023 Source → |
| A written statement of what happens to freed time is scored under both leadership and incentives | The Fear Layer series on this site, part 3 Source → |
| The eight dimensions, the band thresholds (Gap below 40, Weak 40 to 59, Adequate 60 to 79, Strong 80 and above), the tiers and the role targets | Practitioner calibration by Terence Kok. Not drawn from a published standard. |
01 Leadership sponsorship
Whether senior leaders sponsor the change, use the tools themselves and have said in writing what it means for people.
Staff read what leaders do with AI more closely than what they say about it. Sponsorship without visible use reads as a cost-cutting programme.
02 AI literacy by role
Whether each role understands what the AI tools it uses can and cannot do, at the depth that role needs.
Literacy is role-specific. A board member, a line manager and an analyst need different things, and a single all-staff course rarely gives any of them enough.
03 Psychological safety and fear
Whether people can say openly that they use AI, report its errors and raise worries about their jobs without penalty.
Hidden use is the most common sign of fear. People who hide their AI use do not share what works, and they do not report what goes wrong.
04 Job and task redesign
Whether the organisation has mapped the tasks AI changes and updated roles, workloads and targets to match.
AI changes tasks before it changes jobs. When roles and targets stay as they were, people absorb the change as extra work.
05 Verification and automation bias
Whether people check AI output in proportion to the stakes and know where the tools are likely to be wrong.
Fluent, confident output is easy to accept. The risk rises when the tool is usually right, because people stop checking.
06 Incentives and recognition
Whether performance measures, recognition and the handling of saved time reward people for using AI well.
If using AI well only brings more work, or exposes people to scrutiny, the rational choice is to use it quietly or not at all.
07 Training access
Whether people get enough practical, sanctioned time with the tools to build real skill.
A short overview does not change how people work. Practice time on approved tools, ideally with coaching, does.
08 Change capacity
Whether people and managers have the time and energy to absorb another change.
An AI programme competes with every other change running at the same time. Tired teams comply on paper and revert in practice.
Tiers: Fragile 0 to 39; Uneven 40 to 59; Building 60 to 79; Ready 80 to 100.
I built this because the readiness checks I use, my own included, give people a question or two and move on to data. The organisations I have seen struggle with AI rarely had a data problem first. They had a manager who could not answer the question about jobs, or a team that used the tools at home and hid it at work. If your score is low, start with the two dimensions it names and leave the rest for now. Trying to fix all eight at once is how a change programme turns into another source of fatigue.
Last updated 25 Oct 2026
More in Where you stand
- AI Readiness Self-AssessmentAre we ready to deploy AI, and where should we focus first?
- Organisational AI Readiness QuizAcross data, process, technology, people and governance, which one is holding us back?
- AI Value Sequence DiagnosticHave we built the capabilities DBS built, in the order it built them?
- AI Acceptable Use Policy BuilderOur staff already use ChatGPT and Copilot. What should our written rules say?
- AI Use Case Prioritisation MatrixOf the use cases on the table, which do we fund first, and how much autonomy can each one have?
- Business Automation MapWhich of our processes already use AI, and where are the gaps worth filling?
