The Pressure to Have All the Answers About AI

A supplier wants to demonstrate a new system. A colleague asks whether it is safe. Someone else wants to know whether it will change their job. You are still trying to understand what the technology actually does, but because you are the leader, the questions come to you.

Across health and social care, AI is becoming another subject that managers are expected to understand alongside staffing, budgets, quality, safeguarding and the daily demands of running a service. The conversation can move quickly from curiosity to decisions, sometimes before the people involved have had enough time to learn.

On 6 October, the government confirmed that it would accept all 44 recommendations from the National Commission into the Regulation of AI in Healthcare. Its response includes a move towards assessing and monitoring AI-enabled medical devices throughout their working lives, with a full implementation roadmap expected by spring 2027. These are developments in healthcare AI regulation, rather than a blanket approval of every AI product available to services.

For an already busy leader, such announcements can bring a mixture of interest and pressure. There may be opportunities worth exploring, but also an uncomfortable feeling that everyone else understands more than you do. That feeling deserves attention before it pushes you towards pretending to know, avoiding the subject altogether or making decisions too quickly.

You can lead while you are still learning

Leadership can create an expectation of immediate answers. When staff ask whether a tool is reliable or what it means for their work, saying you are still finding out may feel exposing.

Yet an honest answer gives people something dependable. “We have not assessed that yet” is useful information. So is explaining who will examine the question, what evidence is needed and when the team can expect an update.

The difficulty comes when uncertainty is hidden behind confident language. A leader may repeat a supplier’s claims without fully understanding them or offer reassurance about future staffing that has not been agreed. Those statements can be hard to correct later.

Professional confidence can include recognising the limits of your knowledge and arranging the right support. You do not need to become the organisation’s technical expert to make sure important questions are properly examined.

Ask the questions that make the proposal clearer

AI discussions can fill quickly with unfamiliar terms. In a meeting, it may seem easier to nod than interrupt and ask someone to explain what they mean.

Basic questions are often the most useful. What problem is the tool intended to solve? What does it produce? What information does it use? Who checks the result? What happens when it is wrong or unavailable?

A proposal should become clearer as those questions are answered. If the explanation remains difficult to follow, ask for a practical example based on the work your service actually does. A polished demonstration may look impressive while leaving the implications for daily practice unclear.

Different uses also need different expertise. A tool helping draft a general communication raises different questions from one influencing clinical decisions. Leaders should establish the proposed purpose before deciding who needs to assess it.

The Daily Round’s reporting on calls for healthcare systems to act on the risks of autonomous AI offers a related starting point for discussion. The aim of asking about risk is to understand the responsibilities that come with a proposed use.

Staff questions may carry personal worries

When someone asks whether AI will save time, they may also be wondering what will happen to their hours. A question about training may conceal embarrassment about digital skills. Resistance may reflect concern about care quality, previous experiences of poorly managed change or fear that their professional knowledge is becoming less valued.

Leaders should leave room to hear those concerns without assuming what lies behind them. Some staff will be enthusiastic. Others will want more information or need time to consider the implications.

Our article on the quiet fear that AI will make you irrelevant explores the personal side of that uncertainty. Leaders can experience it too, particularly when their own expertise has been built over many years and the language around new technology suggests that everything is changing at once.

Reassurance should remain honest. Explain what is being considered, what has been decided and what remains unknown. Where a proposal could affect roles or working arrangements, those implications need appropriate discussion rather than promises made to settle an uncomfortable conversation.

Share responsibility for understanding the technology

AI should not become an additional subject that one manager is expected to master alone after work.

Depending on the proposal, assessment may need clinical or care expertise, information governance, digital support, procurement, workforce input and the experience of people who will use the system. Smaller providers may need suitable external advice where that expertise is unavailable internally.

Responsibilities should be clear. Someone needs to coordinate the work, but coordination does not mean personally answering every specialist question. It means making sure questions reach the right people and that unresolved issues remain visible.

Staff involvement also needs time. Asking colleagues to explore a tool alongside a full workload can turn “innovation” into another unfinished task. Agree what learning is needed, who will undertake it and how they will share what they find.

How to Introduce AI Without Frightening Your Care Team provides a related read for leaders considering those conversations. Introducing change should include opportunities to question, practise and report difficulties.

Give learning a structure and a boundary

Trying to keep up with every announcement can leave a leader feeling permanently behind. A more manageable approach begins with what is relevant to the service now.

Identify the decisions actually approaching, the knowledge needed to make them and appropriate sources of advice. Set aside time for that learning within the working week where possible. Agree how important updates will be reviewed, so the responsibility does not depend on someone constantly monitoring the news.

There is a useful connection here with the wider question of how professionals develop judgement. In our coverage of Professor Sir Chris Whitty’s call for gradual training in clinical risk, the focus was on developing responsibility through training and supervision. That discussion concerned clinical practice, but it offers a relevant prompt for leaders: people need supported opportunities to build understanding before being expected to carry unfamiliar responsibilities.

A decision can wait for a question to be answered

The pressure to appear forward-thinking can make a pause feel like failure. A supplier may emphasise rapid adoption, while colleagues elsewhere appear to be moving ahead. Neither tells you whether a particular product is appropriate for your service.

Where a question is material to a decision, give it an owner and seek an answer. If an appropriate pilot is agreed, establish its purpose, oversight and review arrangements. Any testing should follow the organisation’s relevant approval processes, particularly where sensitive information or care decisions are involved.

You can remain interested in the possibilities while asking for evidence. You can support learning while acknowledging uncertainty. And you can tell your team that a question needs specialist advice without losing your authority.

The most useful thing a leader brings to an AI conversation may be the willingness to slow it down enough for everyone to understand what is being proposed. Clear questions, shared expertise and honest communication give the team a stronger basis for decisions, and give the leader a more realistic responsibility to carry.