Most boards are approaching AI the wrong way. Not because they’re ignoring it — the opposite. Boards are holding AI strategy sessions, commissioning vendor reports, and adding “AI oversight” to committee charters.

The problem: they’re treating AI as a governance topic when it should be a commercial opportunity.

Here’s what boards are getting wrong — and how to reframe the discussion.

MISTAKE 1: Treating AI as risk management, not value creation

Most board-level AI discussions start with risk:

• What are the regulatory implications?
• How do we mitigate algorithmic bias?
• What’s our data governance framework?

These are important questions. But if that’s where the discussion stays, the board is missing the point.

AI isn’t primarily a risk management topic. It’s a commercial leverage opportunity — and boards should be asking:

• Where can AI compress cycle time or reduce cost-to-serve by 30-50%?
• Which customer pain points could AI solve better than current solutions?
• What proprietary data do we have that could become a competitive moat if we applied AI effectively?

The best boards I’ve worked with flip the sequence: start with commercial opportunity, then layer in governance and risk management to protect the value being created. Risk-first AI discussions lead to slow, cautious implementation.

Opportunity-first discussions lead to disciplined experimentation with clear ROI targets.

MISTAKE 2: Delegating AI to the CTO (or worse, a consultant)

I see this pattern repeatedly: board asks CTO or Chief Data Officer to “lead our AI strategy” — then receives a 40-slide deck full of technology architecture and vendor comparisons.

The problem: AI strategy isn’t a technology question. It’s a business model question.

The critical AI decisions boards need to make aren’t:

• Which LLM provider should we use?
• Should we build or buy AI capability?

The critical decisions are:

• Which parts of our business model could AI fundamentally reshape?
• Are we deploying AI to defend current margins, or to attack new markets?
• What happens to our workforce, our pricing, our customer relationships if AI delivers what vendors promise?

These aren’t CTO questions. These are CEO and board-level strategy questions that require commercial judgment — not technical expertise.

By delegating AI to the technology team, boards abdicate responsibility for the commercial decisions that matter most.

MISTAKE 3: Accepting vendor prmomises without proof

AI vendors are very good at selling the future:
“Our AI will reduce customer support costs by 40%.”

“Our platform will increase sales productivity by 3x.”

“Deployment takes 6-8 weeks with minimal disruption.”

Boards hear these promises, nod approvingly, and approve

budgets. Six months later, the AI project is behind schedule, over

budget, and delivering 20% of promised value.

Why? Because boards aren’t asking the right questions:

• What’s the proof this works in OUR context (not a generic case study)?
• What assumptions underpin the ROI model?
• What needs to be true operationally for this to deliver promised value?
• What’s the kill criterion if it’s not working after 90 days

The best boards I work with treat AI deployment like commercial diligence:

• Pilot first (small, defined scope)
• Measure rigorously (actual results vs vendor promises)
• Scale only after proof (don’t enterprise-deploy on faith)

This isn’t anti-innovation. It’s commercial discipline.

MISTAKE 4: No clear ownership of AI value delivery

Most boards know someone should “own AI strategy.” But when I ask “Who’s accountable for AI delivering commercial value?” — the answer is often vague.

“We have an AI steering committee.”

“The CTO is leading it, with input from the CEO.”

“It’s a cross-functional initiative.”

Translation: no one owns it.

AI projects that succeed have clear ownership:

• One executive (typically CEO, COO, or Chief Commercial Officer) owns the P&L impact of AI deployment
• That executive reports AI progress to the board monthly (not quarterly)
• Success is measured in commercial outcomes (revenue, margin, customer satisfaction) — not AI deployment metrics (models trained, data pipelines built)

If the board can’t name the executive who’ll be fired if AI doesn’t deliver value, the governance structure is broken.

What good looks like: The three questions boards should ask

Instead of “Do we have an AI strategy?” — which inevitably produces a 50-slide deck and no action — boards should ask three specific questions:

1. WHERE HAVE WE RUN A 90-DAY AI PILOT WITH MEASURABLE ROI?

Not “Where are we planning to use AI?” — where have we actually deployed it, measured results, and proven value?

If the answer is “nowhere yet” — the board should ask why. AI is not a 2027 strategy topic. It’s a 2026 execution topic.

2. WHAT’S THE COMMERCIAL CASE FOR SCALING AI BEYOND THE PILOT

If the pilot worked, what’s the business case for scaling across the organization?

This forces the team to connect AI deployment to commercial outcomes:

• If we scale this AI tool across 500 customer support agents, what’s the margin impact?
• If we deploy this in sales, what’s the revenue uplift vs cost of deployment?

If the team can’t build a clear commercial case, don’t scale.

3. WHO OWNS AI VALUE DELIVERY, AND WHAT’S THEIR 90-DAY SCORECARD?

Who’s accountable? What are they measured on? What happens if AI doesn’t deliver?

These questions force clarity and accountability — which is what boards are supposed to provide.

The bottom line

AI is not a governance checkbox. It’s a commercial opportunity that requires the same discipline boards apply to M&A, international expansion, or major product launches:

• Clear business case
• Pilot before scaling
• Measure actual results vs promises
• Assign ownership and accountability

Boards that treat AI this way will capture value. Boards that treat it as a risk management topic will watch competitors move faster.