Do You Need to Be an AI Expert to Lead AI?
Most small businesses do not need a Chief AI Officer or an AI engineer in the boardroom. They do need a named AI owner who understands the business, can ask good questions, set sensible boundaries, involve technical, security and legal specialists, and measure whether AI is making work better.
The UK Government is recruiting senior people to help drive AI strategy and adoption. One recent Civil Service vacancy attracted attention because previous experience in digital, data, innovation or AI was listed as desirable rather than essential for a strategy and engagement role.
That prompted an understandable reaction: how can someone lead AI without being an AI expert? But the question may confuse AI expertise with AI leadership. Would you expect the CEO of a construction company to be the best bricklayer? Would you expect the person setting Microsoft 365 strategy to be the best PowerShell engineer? Probably not. The same distinction increasingly applies to AI.
Quick answer
No — not necessarily. You do not need to know how to build AI to lead its adoption. You do need to understand enough to know what good looks like.
An effective AI leader combines business understanding, AI literacy, authority to make decisions and access to technical, security and legal expertise. They do not need every answer. They need ownership of finding the right answer before the business takes the risk.
AI leadership without AI understanding is not enough. “I am not technical” is not a governance model.
AI technical expertise and AI leadership are different capabilities
An AI technical specialist and an AI or business leader may work closely together, but they are answering different questions. The specialist may be responsible for making a system work. The leader is responsible for deciding whether it should exist, what success means and what the business is prepared to accept.
| AI technical specialist | AI or business leader |
|---|---|
| Models, tools, APIs and integrations | Business problems, priorities and outcomes |
| Data flows, architecture and implementation | People, workflow, change and adoption |
| Testing, monitoring and technical controls | Governance, approval and accountability |
| What the system can do and how it can be built | Why the organisation should do it and for whom |
The question that keeps AI useful
The AI specialist answers “how?”. The AI leader must also answer “why?”.
A strong AI programme usually needs both capabilities. Separating them does not make technical expertise less important. It prevents the technical team from being left to invent the business case, employee policy or acceptable level of risk on its own.
The real SME need: an AI owner
Most SMEs do not need another C-suite title. They do need someone to own the question. That person may be the business owner, operations lead, IT lead, transformation lead, finance leader, technically confident manager or another senior person with cross-business visibility.
The important point is not the job title. It is that someone is accountable for understanding how AI is being used and for moving decisions forward. Without an owner, AI becomes a collection of individual subscriptions, experiments and assumptions.
Small businesses probably do not need another C-suite title
They do need someone to own the question:
- Where are we already using AI?
- Which teams are using it, and which tools are involved?
- What data goes into those tools?
- What problems are we trying to solve?
- Which use cases are approved?
- What is working and what is risky?
- What should we test next?
- Who needs technical, security or legal help?
Ownership is about accountability, not enthusiasm or job title. The person who used ChatGPT first is not automatically the right person to own company data, automation, AI agents, customer communications, employment decisions or supplier risk. Equally, the most senior IT engineer is not automatically the best AI owner if they do not have cross-business authority.
The first question: is your business already using AI?
This sounds easier than it is. Management may know that Sarah uses ChatGPT, that the company has Copilot or that marketing uses Canva AI. Employees may also be using ChatGPT, Gemini, Claude, Copilot, AI built into a CRM, meeting transcription, image tools, browser assistants and SaaS features without central visibility.
You cannot lead AI adoption if you do not know where AI adoption has already happened.
The absence of an AI programme does not mean the absence of AI. It may mean the business has not yet asked its people, reviewed its licences or looked at the features already present in software it pays for.
Shadow AI is a visibility problem before it is a technology problem
Shadow AI is AI being used outside established business visibility, policy or control. It is not automatically malicious, and it is not a reason to shame employees for finding a useful tool. It is a sign that the business needs a safer, clearer way to discuss and support AI use.
Unseen use can create risks involving confidentiality, customer information, intellectual property, security, personal data and inconsistent output. The sensible response is to discover what is happening, understand the actual use case, set data boundaries and provide an approved alternative where one is needed.
Read: Shadow AI — The Tools Your Team Isn't Telling You About →
What should the AI owner be good at?
There is no single correct background. The right person should ideally have six qualities:
- 1Credibility — people will listen to them and take the work seriously.
- 2Business understanding — they know how work actually gets done, including the awkward exceptions in the process map.
- 3Curiosity — they are willing to learn enough about AI to ask useful questions and test assumptions.
- 4Scepticism — they can challenge hype, weak business cases and unsafe shortcuts.
- 5Authority — they can approve, pause or stop an initiative rather than merely write a recommendation.
- 6Access to expertise — they know when technical, security, privacy, legal or employment advice is required.
The AI owner does not need every answer. They need ownership of finding the answer, recording the decision and making sure the right people are involved before an experiment quietly becomes a production dependency.
AI literacy, not necessarily AI engineering
An AI owner should understand enough to explain hallucinations, model limitations, data privacy, AI agents, automation, permissions, vendor risk, Shadow AI and human oversight. They should know the difference between a system making a recommendation and a system taking an action.
They do not necessarily need to train models, write Python, build APIs or manage GPU infrastructure. Those may be specialist responsibilities. But an owner who cannot ask what data is used, what the system can do, who checks important outputs or how it can be switched off is not ready to increase AI autonomy.
The 80/20 AI knowledge test
Before increasing the scope of an AI tool or agent, the owner should be able to answer:
- What AI tools are we using?
- What data are we entering?
- Can the supplier use our data for training, and what do the terms say?
- What can the AI actually do?
- What can it not reliably do?
- Does it make recommendations or take actions?
- Who checks important outputs?
- What happens when it is wrong?
- Who can switch it off?
- How are results measured?
If the owner cannot answer these questions, the business needs support before it gives the system more data, more permissions or more autonomy.
The expertise gap: where specialists still matter
AI leadership without specialists can fail. A business leader may correctly identify invoice processing as a worthwhile problem, then need specialist input on architecture, security, permissions, integration, data protection, testing and monitoring.
Technical experts should normally remain responsible for model or API integration, identity, security architecture, data flows, permissions, infrastructure, application controls, logging, agent permissions and technical testing. This article is not suggesting that a non-technical leader should configure those controls themselves.
The leader owns the outcome. Specialists help make the outcome technically safe and achievable. Where an AI system handles personal data, employment decisions, regulated work, confidential information or consequential customer outcomes, involve the relevant privacy, legal, security or specialist adviser before deployment.
Read: Your AI Agent Did Something Illegal. Who Is Responsible? →
AI leadership is change management
Many technology projects do not fail because the software could not do the thing. They fail because people did not use it, the problem was poorly defined, the workflow did not change, management did not measure value or nobody owned adoption. AI magnifies those weaknesses because it is easy to buy and easy to experiment with.
Buying AI is easy. Changing how the business works is the harder part.
AI owners will have to work with enthusiasts, cautious adopters, sceptics, people worried about jobs and staff who are already using AI informally. They need to create a route for questions and feedback, not just issue a prohibition that pushes useful or risky use underground.
Read: Are You an AI Zoomer, Bloomer, Gloomer or Doomer? →
Do not make AI only an IT project
IT should be heavily involved in security, access, data, integration and technical controls. But IT cannot decide whether sales, customer service, finance or operations has the best use case. Those decisions depend on business priorities, customer commitments, process knowledge and the people who will have to change how they work.
The ownership split
IT can enable AI. The business has to decide what problem AI is supposed to solve.
Do not make AI only a marketing project either. Marketing may be an early adopter, but AI can affect finance, HR, customer service, operations, IT, management, sales and compliance. The owner needs cross-business visibility.
A proportionate AI steering group
A larger SME may benefit from a small cross-functional group: the AI owner, an IT or security representative, an operations representative and business users from the teams affected by the use case. The group should make decisions and remove obstacles, not create a new committee that meets without changing anything.
For a ten-person business, one accountable owner and a periodic management review may be enough. Larger organisations may need a Chief AI Officer where AI is core to the product, investment is substantial, regulation is significant, large AI engineering teams exist or AI is a strategic differentiator.
Governance should be proportional to the business and the consequence of the use case, not proportional to how fashionable the technology is.
You probably need AI ownership before an AI job title
A Chief AI Officer can be a legitimate role in a large organisation. The title is not the problem. The mistake is assuming that a title alone creates a strategy, or that a small business needs a senior appointment before it has mapped its current use of AI.
For most SMEs, start by naming an owner, agreeing their authority, giving them time to learn and connecting them to the technical, security, privacy and legal help they may need. If the scope later justifies a dedicated role, the business will have better evidence for defining it.
The IT Club AI Owner model
Use this sequence to turn AI interest into accountable adoption:
| Stage | Question to answer | Practical output |
|---|---|---|
| Discover | What AI is already being used? | A simple inventory of tools, users, data and current purpose |
| Prioritise | Which business problems are worth solving? | A short use-case pipeline tied to business outcomes |
| Control | What data, tools and actions need boundaries? | Approved use, data rules, permissions and human-approval points |
| Enable | How can staff use AI safely and consistently? | Approved tools, guidance, training and a route for questions |
| Measure | Is it improving work, saving time or reducing cost? | Baseline measures, user feedback and evidence of value |
| Review | What changed and what needs to change next? | A decision to continue, adapt, pause or stop the use case |
A useful starting principle
AI ownership starts with visibility, not procurement.
The first action may be a conversation with staff and a review of existing software licences, not another AI subscription.
AI ownership checklist
Use this as a conversation starter rather than a fake score. The result is a description of your current position, not a percentage that pretends to measure readiness.
Management check-in
Who owns your AI decisions?
A short, plain-English conversation starter for UK SMEs. It ends with one practical next step, not a lead form.
- Named AI owner
- AI tool inventory
- Approved and unapproved tools understood
- AI usage policy or practical rules
- Data boundaries
- Technical and security support identified
- Use-case pipeline
- Human-approval rules for consequential outputs or actions
- Measurement of value and quality
- Staff feedback route
- Review cycle
| Current position | What it means |
|---|---|
| Clear ownership | A named person has authority, visibility and access to expertise. |
| Partial ownership | Some controls or experiments exist, but responsibility or coverage is incomplete. |
| No clear owner | AI use is being left to individual teams, suppliers or chance. |
Ten questions for management
- 1Who owns AI here?
- 2What AI are we already using?
- 3Which use cases matter to the business?
- 4What is prohibited?
- 5Where does business data go?
- 6Who evaluates new tools and supplier terms?
- 7Who approves automation?
- 8Who monitors AI agents or other systems that can take actions?
- 9Who measures value and quality?
- 10Who reviews all this again?
If the answer to every question is “IT”, the business probably has not actually created an AI strategy.
The Operational Heartbeat
AI ownership cannot be a one-off workshop. Review new tools, existing AI use, Shadow AI, approved applications, data exposure, new agent capabilities, supplier changes, staff adoption, value delivered, incidents and failed experiments on a recurring basis.
AI needs an Operational Heartbeat: tools, use cases, data, controls, adoption and outcomes should be reviewed because AI capability changes far faster than most business policies.
The IT Club view
You do not need your AI leader to be the person who knows the most about large language models. You need them to be the person accountable for AI making the business better without creating uncontrolled risk.
The four things your AI owner needs
Business understanding + AI literacy + authority + access to expertise.
The best AI owner may be a translator between the people who understand the technology and the people who understand the business.
For most SMEs, the next step is not recruiting a Chief AI Officer. It is naming somebody who owns AI adoption, giving them enough understanding to ask the right questions and making sure they have access to specialist help when the questions become technical.
Frequently Asked Questions
Do you need AI experience to lead AI?
You do not necessarily need AI engineering experience. You do need AI literacy: enough understanding of limitations, data, permissions, automation, human oversight and supplier risk to make responsible decisions and know when to involve a specialist.
Do small businesses need a Chief AI Officer?
Usually not as a first step. Most small businesses need a named AI owner with authority, time and access to expertise. A dedicated Chief AI Officer may make sense when AI is central to the product, investment and risk are substantial, or the organisation has a large AI function.
Who should own AI in a small business?
There is no universal answer. It could be the owner, operations lead, IT lead, transformation lead, finance leader or another senior person with cross-business visibility. Choose the person who can understand the business, challenge weak ideas, approve or stop work and bring in specialist help.
Should IT own AI?
IT should own or heavily support technical controls, security, access, data flows and integrations. The business should own the reason for using AI, the outcomes it wants and the people and processes that need to change. One team should not be expected to own both sides alone.
What is Shadow AI?
Shadow AI is AI used outside established business visibility, policy or control. It may be a useful experiment, but it can also expose confidential, personal or customer information. Discovering it and offering a safe approved route is usually more effective than pretending it does not exist.
How should an SME start an AI strategy?
Name an owner, discover current use, prioritise a small number of worthwhile problems, set data and approval boundaries, enable staff with approved tools, measure outcomes and review the decision. Start with visibility and a manageable use case rather than buying technology first.
Sources and further reading
The news hook for this article was a Telegraph report about a government AI-related appointment. The report was not treated as evidence that technical knowledge is unimportant, and its headline was not repeated as a fact. The useful question is what a strategy and engagement role actually needs to achieve.
The Telegraph — Government seeks AI chief with no AI experience needed →
Civil Service Jobs — Head of Strategy & Engagement (role listing) →
GOV.UK — AI Opportunities Action Plan: government response →
Read: The Small Business AI Mistake That Looks Like Progress →
Plain-English Takeaway
You do not need to know how to build AI to lead its adoption. You do need to understand enough to know what good looks like — and you need authority to make the business act on that understanding.
Tap to follow The IT Club Briefing on WhatsApp.
Related Articles
Are You an AI Zoomer, Bloomer, Gloomer or Doomer?
Four broad attitudes towards AI are useful in a business meeting: move faster, test it properly, show me the risks, or question whether we should use it at all. Here is how an SME can use all four viewpoints.
Read articleGoogle Maps Is Becoming an AI Agent — Why Local Businesses Should Care
Google Maps is evolving from a place-search tool into a conversational AI assistant that can help users find businesses and, in supported situations, move towards ordering, booking or reserving. The shift matters because an AI agent may narrow the customer's choices before they ever see the map.
Read articleThe Race to Replace: Why AI Needs to Start Doing the Work
Generative AI started by answering questions. The bigger economic shift may come when AI agents take responsibility for pieces of real work — and jobs are bundles of tasks, not indivisible units.
Read article