10 AI Questions Every Board Should Be Asking Management
Artificial intelligence is becoming a Board issue.
Not because directors need to understand every AI model, evaluate individual tools or approve every use case.
They don't.
But AI increasingly affects areas Boards are already responsible for overseeing:
Strategy.
Investment.
Risk.
People.
Technology.
Customers.
Operational resilience.
Competitive position.
And business performance.
That means Board oversight of AI needs to move beyond occasional updates about pilots, licences and new technology.
The more useful question is:
Does the Board have enough visibility to determine whether management is making good decisions about AI?
For many organisations, the answer will depend less on how much directors know about AI technology and more on the quality of the questions they ask management.
Here are ten questions every Board should be asking.
What is the Board's role in AI?
The Board's role is oversight, challenge and assurance.
Management remains responsible for running the organisation and executing its AI strategy.
The Board should not become an AI steering committee.
But directors should seek sufficient confidence that:
AI supports the business strategy;
material opportunities are being identified;
investment is appropriately prioritised;
accountability is clear;
significant risks are governed;
workforce implications are understood;
material AI activity is visible;
value is being measured.
This is an important distinction.
Good Board oversight does not mean more involvement in operational AI decisions.
It means better visibility of the decisions that matter.
For a broader examination of director responsibilities, read Board Oversight of AI: What Directors Actually Need to Know
Question 1: Where are we actually using AI?
This should be one of the simplest questions.
It can also be surprisingly difficult to answer.
AI may already exist across the organisation through:
formally approved AI platforms;
generative AI tools;
software with embedded AI features;
departmental applications;
automation;
employee experimentation;
customer-facing systems;
AI agents.
Some of this may be visible centrally.
Some may not.
This creates the possibility of a significant gap between:
where leadership believes AI is being used
and
where AI is actually being used.
Boards don't need an inventory of every employee interaction with AI.
But they should seek confidence that management has sufficient visibility of material AI use.
A useful follow-up question is:
Which AI use exists today that management would be uncomfortable explaining to the Board?
That can expose shadow AI and weak organisational visibility quickly.
For more on this risk, read Shadow AI and AI Governance: The Hidden Risk of Uncontrolled AI Use in Your Business
Question 2: How does AI support our business strategy?
AI should not exist as a strategy alongside the business strategy.
The Board should be able to see the connection between AI investment and organisational priorities.
If the strategy requires the business to:
increase operating margin
how will AI contribute?
If the priority is:
revenue growth
where could AI materially support it?
If the organisation needs:
better customer retention
which AI capabilities are relevant?
If the business wants:
growth without equivalent headcount growth
where can AI create capacity?
This shifts the conversation from:
“What AI are we implementing?”
to:
“Which strategic outcomes are we using AI to improve?”
If management cannot make that connection clearly, AI may be developing as a technology programme rather than a business strategy.
Question 3: Which AI opportunities are we deliberately NOT pursuing?
This is an unusually revealing Board question.
Management teams can usually produce a list of AI projects.
But strategy requires choices.
If every possible AI opportunity is considered a priority, the organisation does not really have priorities.
The Board should understand:
What have we evaluated?
What have we prioritised?
What have we deferred?
What have we rejected?
Why?
A healthy AI portfolio should contain no decisions.
Not every AI opportunity deserves investment.
The ability to reject an interesting technology because the business case is weak is a sign of strategic discipline.
Question 4: Who is accountable for AI?
The answer should not simply be:
“The CIO.”
Nor:
“The AI Committee.”
AI may involve Technology, Risk, Legal, HR, Data, Finance and individual business functions.
But the Board should understand where accountability ultimately sits.
Ask:
Who has executive accountability for our overall approach to AI?
Then:
Who owns each material AI use case and its business outcome?
These are different questions.
Technology might own the system.
A business leader should normally own the outcome the system is intended to create.
If everyone is involved but nobody is clearly accountable, the organisation has an accountability gap.
Question 5: What are our most significant AI risks?
The Board does not need a list of every theoretical AI risk.
It needs visibility of material exposure.
Depending on the organisation, that might include:
confidential data;
personal information;
cybersecurity;
inaccurate outputs;
bias;
customer outcomes;
regulatory exposure;
intellectual property;
operational resilience;
third-party dependency;
autonomous actions;
reputation.
The Board should also understand whether management is applying proportionate governance.
If every AI use case receives the same controls, governance may either be too weak for high-risk applications or unnecessarily restrictive for low-risk ones.
The better question is:
Do our strongest controls sit around our most consequential AI uses?
For the wider framework, read What Is AI Governance? A Practical Guide for Business Leaders
Question 6: Where does AI require meaningful human oversight?
Boards will increasingly hear:
“There is a human in the loop.”
That statement alone should not provide assurance.
Directors should ask:
What is the human actually responsible for?
Can they understand the AI output?
Can they challenge it?
Can they override it?
Do they have enough time to review it properly?
What happens if they disagree with the AI?
A nominal human reviewer can create an illusion of control.
Meaningful human oversight requires genuine authority and practical ability to intervene.
This becomes particularly important where AI influences significant decisions affecting employees, customers, finances or other stakeholders.
Question 7: How much are we really spending on AI?
This question may produce a surprisingly complicated answer.
AI expenditure can be spread across:
enterprise platforms
software licences
embedded AI features
cloud/model usage
consultancy
data
integration
training
security
governance
departmental budgets
and
existing technology contracts.
The Board does not need every invoice.
It does need sufficient visibility to determine whether AI investment is being managed as a portfolio.
Ask:
What is our total material AI investment?
Which initiatives consume the most capital?
Which outcomes are those investments intended to create?
Who owns those outcomes?
What evidence supports continued investment?
That moves AI from technology expenditure into capital allocation.
Question 8: What measurable value has AI created?
This may be the most important question on the list.
Management reporting can easily focus on activity:
number of AI licences
number of users
number of pilots
number of employees trained
number of hours theoretically saved
But Boards ultimately need to understand outcomes.
Ask:
What changed in the business?
Did AI:
increase revenue?
reduce cost?
create capacity?
improve customer outcomes?
accelerate processes?
reduce errors?
improve quality?
reduce risk?
Then:
What did it cost to achieve that outcome?
This distinction matters because theoretical productivity is not automatically financial value.
Question 9: Are our people actually ready for the AI strategy?
AI transformation depends on employees and managers changing how work gets done.
The Board should therefore challenge assumptions about adoption.
Ask:
Are employees actually using the AI capabilities we've purchased?
Do they understand acceptable use?
Do they have the skills required?
Which groups have low adoption?
How are roles changing?
What concerns are employees raising?
Are managers equipped to lead AI-enabled teams?
This moves workforce reporting beyond:
“95% completed AI training.”
Training completion is an input.
It does not prove adoption.
Question 10: What should we scale, change or stop?
Boards should be cautious if every AI initiative is permanently described as successful.
Experimentation should create evidence.
Evidence should create decisions.
Some initiatives should:
SCALE
because the evidence supports further investment.
Others should:
CHANGE
because the opportunity remains attractive but the current approach is underperforming.
Others should:
STOP
because achievable value no longer justifies the cost, risk or organisational effort.
Stopping an AI initiative is not necessarily failure.
Continuing to invest after the evidence no longer supports the business case can be.
The Board should therefore ask management:
Which AI initiative have we stopped recently—and what did we learn from it?
If the answer is none, it is worth understanding why.
The Stratify Board AI Oversight Model
These ten questions can be grouped into five areas of Board oversight:
DIRECTION
Where are we using AI and how does it support strategy?
OWNERSHIP
Who is accountable for decisions and outcomes?
CONTROL
What are the material risks and where is human oversight required?
INVESTMENT
What are we spending and why?
VALUE
What is creating measurable value and what should scale, change or stop?
That gives the Board a simple oversight sequence:
DIRECTION → OWNERSHIP → CONTROL → INVESTMENT → VALUE
The Board does not need to operate the AI portfolio.
It needs sufficient visibility across these five areas to challenge whether management is operating it effectively.
What should an AI Board report contain?
Boards do not need a 60-page AI pack.
A useful report might include:
Strategic position
Major developments affecting business strategy.
Material AI portfolio
The most important initiatives and their owners.
Investment
Current and planned material expenditure.
Value
Expected versus realised outcomes.
Risk
Highest-risk use cases and significant changes.
Adoption
Meaningful workforce or customer adoption issues.
Incidents
Material exceptions, failures or governance issues.
Decisions
Items requiring Board awareness, challenge or approval.
The objective is not more reporting.
It is better Board decisions.
How often should Boards discuss AI?
There is no universal schedule.
For organisations where AI is strategically significant, it should increasingly become part of normal Board discussions rather than an annual technology presentation.
AI may arise naturally during:
strategy
risk
investment
workforce
technology
customer
and
performance
discussions.
That is ultimately a sign of maturity.
AI stops being treated as a separate novelty and becomes part of how the organisation is governed.
What Boards should avoid
There are two extremes.
The passive Board
“Management understands AI. We'll leave it to them.”
This risks insufficient challenge of an increasingly material business issue.
The operational Board
“We want to approve individual AI tools and projects.”
This pulls directors into management.
The better position is informed oversight.
Understand enough to ask good questions.
Require evidence.
Challenge assumptions.
Seek clarity of accountability.
And focus on material decisions and outcomes.
AI literacy for directors is really decision literacy
Boards do not need every director to understand transformer architecture, model parameters or the technical differences between every AI platform.
They need enough understanding to recognise:
opportunity
risk
weak assumptions
unclear accountability
poor investment discipline
and
missing evidence.
The objective of Board AI literacy is not technical expertise.
It is better governance and better decisions.
The Board AI Test
At your next Board meeting, ask whether directors can answer:
Where is AI materially used?
How does it support strategy?
What aren't we pursuing?
Who is accountable?
What are our biggest risks?
Where is human oversight required?
How much are we investing?
What value have we realised?
Is the workforce adapting?
What should scale, change or stop?
If several answers depend on:
“We'll need to ask IT.”
the Board may not yet have sufficient visibility.
The solution isn't necessarily another Board paper.
It may be a better AI management system underneath the Board reporting.
Because good oversight depends on good information.
And good information depends on clear ownership, governance and measurement throughout the organisation.
Frequently Asked Questions
What should Boards know about AI?
Boards should understand how AI relates to business strategy, where it is materially used, who is accountable, significant risks, major investment, workforce implications and whether AI is producing measurable value.
What questions should Boards ask about AI?
Boards should ask where AI is used, how it supports strategy, who owns it, what risks exist, where human oversight is required, how much is being invested, what value has been realised and which initiatives should scale or stop.
Is AI a Board responsibility?
Boards are responsible for appropriate oversight of material AI opportunities and risks. Executive management remains responsible for strategy execution and day-to-day AI governance.
Does a Board need an AI expert?
Not every Board necessarily needs a dedicated AI specialist. Directors collectively need sufficient AI literacy to challenge management effectively, with access to specialist expertise where required.
Should Boards approve AI projects?
Boards generally should not approve routine AI projects unless existing governance or investment thresholds require it. Their focus should remain on material strategy, investment, risk and oversight.
How often should the Board review AI?
The appropriate frequency depends on how material AI is to the organisation. Where AI is strategically significant, it should increasingly feature within regular strategy, risk, investment and performance discussions.
What should an AI Board dashboard include?
A useful Board view can include major AI initiatives, business owners, investment, expected and realised value, material risks, adoption, significant incidents and decisions requiring Board attention.
How Strong Is Your Organisation's AI Governance?
Boards can only oversee what management can see.
Stratify's Executive AI Governance & Risk Assessment helps leadership teams examine whether the foundations required for responsible AI adoption are actually in place.
Assess areas including:
Accountability
Governance foundations
Human oversight
Data, security & access
Risk and controls
Monitoring and incident response
Test Your AI Governance
Because Board oversight shouldn't begin with:
“Tell us what AI you're doing.”
It should begin with:
“Show us how you know you're making the right AI decisions.”
Comments