My leadership team wants to ignore AI. How do I get them involved?

The talk was about a leadership problem that is becoming harder to avoid. AI can now summarise information, draft arguments, review material and suggest options. It can sit quietly inside a decision process and shape what people see, how they think and what they recommend. But AI cannot carry responsibility.

That was the point behind one of the best questions asked after the session: “My leadership team wants to ignore AI. How do I get them involved?”

It is a good question because it reflects a problem many organisations now face. Some teams are experimenting with AI. Some leaders are enthusiastic. Others are cautious, sceptical or hoping the issue will stay with IT. Ignoring AI may feel like the safe option but it’s not.

Ignoring AI does not remove the risk

If a leadership team chooses not to engage with AI, that does not mean AI is absent from the organisation. People may already be using public tools to draft reports, summarise documents, prepare briefings, analyse information or sense-check ideas. Some of that use may be harmless. Some may be useful. Some may create risks around confidentiality, accuracy, judgement and accountability.

The risk is not only that AI makes a mistake. The deeper risk is that nobody knows where it is being used, what information is being put into it, what decisions it is influencing, and where human judgement is being weakened.

Research on AI adoption shows that generative AI has spread quickly. Bick, Blandin and Deming reported that, as of late 2024, nearly 40% of the US population aged 18 to 64 used generative AI, with 23% of employed respondents using it for work at least once in the previous week.

That does not mean every organisation is using AI well. It means leadership teams should not assume that non-engagement equals non-use.

Do not start by asking leaders to believe in AI

The worst way to involve a reluctant leadership team is to ask them to become excited about AI in general. Most leadership teams do not need a lecture on artificial intelligence. They need to understand how it may affect the work they are already responsible for.

A better starting point is to choose one familiar area of work and ask:

• where does the information come from?
• who checks it?
• where could AI already be involved?
• what would happen if the AI output was wrong?
• who remains accountable for the decision?

That might apply to a board note, an incident report, a risk assessment, a crisis update, a policy review or an executive briefing.

The practical leadership issue is not whether AI is impressive. It is whether leaders understand where it is shaping information, judgement and action.

The danger is not only hallucination

Most people now understand that AI can hallucinate. It can produce confident answers that are wrong, incomplete or poorly grounded. That matters, but it is not the only issue.

A more subtle danger is that a fluent answer can make people less willing to do the hard work of judgement. Steven D. Shaw and Gideon Nave have described this as “cognitive surrender”: the tendency to defer judgement, effort and responsibility to AI output, especially when the system appears capable and confident.  That is a leadership problem, not just a technical one.

If a junior person uses AI to draft a report, who checks the assumptions? If a manager uses AI to summarise a complex issue, who checks what was missed? If a leadership team uses AI to frame a strategic choice, who challenges the recommendation?

The risk is not that AI has entered the room. The risk is that people stop noticing how much authority they are giving it.

Productivity claims should be tested, not assumed

Another reason leadership teams need to engage is that AI claims are often too broad. AI may save time in some settings. It may improve quality in some settings. It may reduce effort, speed up drafting, support analysis or help people find information faster. But those benefits need to be tested against real work.

The “No Man’s Hand” study on AI-assisted police report writing is a useful warning. The study found that AI-assisted report writing did not reduce the time officers spent writing reports, despite the expectation that the technology would speed up the process.

That does not mean AI is not useful. It means leaders should be careful with simple claims such as “AI will save time”.

In operational settings, the real value may not be fewer minutes spent typing. It may be better review, clearer evidence, fewer missed issues, more consistent briefing or stronger follow-up. Those are different claims. They need different evidence.

AI should be treated as normal technology

One of the most useful ways to reduce fear around AI is to stop treating it as magic. AI is powerful, but it is still technology. Its impact depends on where it is used, what it is connected to, what people rely on it for, and what checks sit around it.

That is why I find the idea of AI as “normal technology” useful. It moves the discussion away from extremes. AI is not an autonomous species taking over the organisation. It is also not a harmless writing assistant that can be ignored.

It is a general-purpose technology that will change habits, expectations and ways of working over time.

Leadership teams need to engage because the organisation’s habits will change whether or not they formally approve a strategy. The choice is whether that change is visible and managed, or informal and hidden.

Start with one leadership conversation

If a leadership team wants to ignore AI, do not begin with a large transformation plan. Start with a practical conversation. A useful first meeting could cover five questions:

1. Where are people already using AI, formally or informally?
   This gives leaders a realistic view of current use, rather than relying on policy assumptions.
2. Which decisions or outputs must remain clearly human-owned?
   This is especially important where safety, security, resilience, legal, financial or reputational consequences are involved.
3. Where would a confident but wrong AI answer cause harm?
   This helps identify the areas where review, evidence and escalation matter most.
4. Which one task could be tested safely with controlled material?
   Examples might include summarising reports, reviewing written material, finding internal guidance, preparing briefings or analysing recurring issues.
5. What evidence would we need before allowing wider use?
   This moves the discussion from opinion to measurement.

That conversation is more useful than asking whether people are “pro-AI” or “anti-AI”. It brings the issue back to responsibility.

Leadership means preventing assistance from becoming dependency

The challenge for leaders is not to block AI or accept it uncritically. The challenge is to keep human judgement visible.

That requires new habits. Leaders need to ask where AI is being used, what source material it relied on, what was checked, what was uncertain, and who remains responsible for the final decision.

This matters most in higher-consequence work. In safety, security, resilience, risk and crisis settings, AI may help teams find information, prepare briefings, review material and understand a situation more quickly. But the responsibility for judgement, action and consequence remains human. That is where leadership has to sit.

When AI can inform decisions but not carry responsibility, leadership means preventing assistance from becoming dependency.

Author bio: Andrew Tollinton

Andrew Tollinton is CEO and Co-Founder of SIRV, which builds operational AI for safety, security and resilience teams. He focuses on practical, controlled AI use in serious environments, with particular interest in evidence, accountability and human judgement. Andrew chairs the Institute of Strategic Risk Management’s AI in Risk Management Special Interest Group and speaks regularly on AI governance and operational resilience.