Why the leadership challenge behind AI is not speed, but judgement

AI saves me around 40% of my working time.

It has not removed 40% of my responsibility.

I have used AI consistently, virtually every day, for around three years. As the owner of a small business, I regard it as an extraordinary advantage. It completes administrative work, helps me organise information, supports my content production and carries out tasks that would previously have consumed a significant part of my week.

I can point it towards something, explain what I need and allow it to do much of the execution on my behalf.

But I still make the decisions.

That distinction is becoming increasingly important.

Because although AI can produce something impressive in seconds, it can also produce something completely wrong with the same confidence. Even after three years of daily use, I remain surprised by some of the errors it makes.

The technology saves me time, but it also demands vigilance. I have to check the information, challenge the assumptions and make sure an apparently polished answer is not disguising a serious mistake.

That is the leadership challenge behind AI.

It gives us more speed, more output and greater access to information.

But faster does not automatically mean better.

AI can help leaders move faster. It cannot decide what is worth moving towards.

The speed is real. So is the responsibility.

I am not anti-AI. My own business is evidence of its value.

A small business can now access capabilities that once required several members of staff, specialist agencies or significant administrative support. Tasks can be researched, structured, categorised and completed much more quickly.

Used properly, this creates time.

The question is what we do with that time.

Do we use it to think more deeply, improve our decisions and create space for the work that really matters?

Or do we immediately fill it with more output, more communication, more projects and more activity?

That is where high achievers may be particularly exposed.

We are conditioned to use available capacity. An empty hour in the diary can feel like a problem to solve. If AI helps us complete ten tasks more quickly, the temptation is to add another ten.

The diary remains full. The volume increases. The organisation accelerates.

But nothing necessarily becomes clearer.

I explored a similar pattern in Where Ambition Outpaces Execution. A full diary proves activity, but it does not prove that the right work is being done.

AI has not created that leadership problem.

It has magnified it.

AI can provide an answer. It cannot carry accountability.

Personally, I do not regard the current generation of AI as intelligence in the same sense that we use the word about a human being.

It is extraordinarily powerful at processing information, recognising patterns and producing responses from vast bodies of available material. It can work through instructions and apply rules quickly.

For an administrative process, those rules might be straightforward:

• If this condition is met, place the item in one category.
• If another condition is met, place it somewhere else.
• If the available information is insufficient, mark it as unknown and return it for review.

That can be immensely useful. But somebody still created the criteria. Somebody decided what the categories meant, and somebody has to consider the consequences of getting it wrong.

The person delegating the task remains accountable for the result.

That matters even more when decisions involve people. AI does not bring lived experience, emotional intelligence, personal values or genuine responsibility for what happens afterwards. Medical commentators have similarly argued that AI may perform particular tasks extremely well without possessing the contextual judgement associated with human cognition.

Harvard Business Review has warned against assuming that more data and more powerful algorithms will automatically uncover the truth and create better decisions. Human strengths still differentiate judgement, particularly when situations contain ambiguity, competing values and incomplete information.

The leader can delegate an action.

The leader cannot delegate accountability.

I know what capability without restraint can cost

I learned this long before generative AI appeared.

In late 2017, I took responsibility for a failing business with approximately 15 employees and turnover of around $6 million.

I could see what was possible.

I believed the business could recover, grow and become something considerably stronger. I brought high standards, pace, energy and a relentless focus on improvement.

The approach worked.

The business stabilised. Over the following years, turnover roughly doubled to $12 million and the team grew to around 30 people.

From the outside, it looked like success.

In many respects, it was.

But I nearly broke myself in the process. I came close to damaging my personal life, and I placed a level of pressure on a previously fatigued team that they could not sustainably carry.

My capability had become part of the problem, and because I could work at that pace, I continued.

Because I could see what was possible, I kept pushing, and the results were improving, it was easy to treat the cost as necessary and temporary. It was addictive.

I wrote more about that experience in I Hit the Wall: What High Achievers Don’t Admit About Burnout and Performance.

The lesson was not that ambition, speed or capability are bad.

It was that capability without clarity can take a person, a team or an organisation a long way before the damage becomes visible.

AI places more capability into more hands.

That is exciting. It is also why judgement and restraint matter.

More information does not always create better conversations

The implications differ across professions.

Doctors increasingly meet patients who have used AI to research their symptoms, interpret results or prepare pages of possible diagnoses before an appointment. Lawyers can face clients armed with AI-generated interpretations of legislation, contracts or precedent.

Better-informed patients and clients can be positive.

But information without context can also create false certainty.

The professional may have to spend valuable time correcting inaccurate information before the real conversation can begin. A polished document does not necessarily reflect clinical understanding, legal training or an appreciation of risk.

The BMJ has reported that patients increasingly use chatbots to prepare for appointments and interpret symptoms. It argues that clinicians need stronger AI literacy, critical integration skills and communication capabilities to manage this safely.

Research published through The Lancet’s EClinicalMedicine has also argued that clinical AI requires meaningful clinician involvement and human oversight throughout its development and use. The technology must be matched to the context in which decisions are actually made.

The same principle applies beyond medicine.

Information is not judgement.

Access is not expertise.

And confidence is not accuracy.

Larger organisations face a different clarity problem

In one organisation of several hundred people, I saw a small group begin exploring AI for sensible reasons.

They understood what it could do and wanted to move forward.

However, perhaps 80–85% of the wider organisation had limited understanding of AI, its potential or its risks.

That created difficult leadership questions.

• Should the informed minority be allowed to move quickly?

• Which systems and information could they use?

• What restrictions were necessary?

• How could sensitive data be protected?

• Who was responsible when AI-generated work influenced a decision?

• How could the organisation avoid suppressing useful experimentation without creating an uncontrolled free-for-all?

This is why AI adoption can become a rabbit hole.

The outcomes may be enormous, but the journey towards them requires governance, education and clear decision rights.

NIST’s Generative AI Profile recommends that organisations incorporate trustworthiness and risk considerations throughout the design, use and evaluation of AI systems rather than treating governance as something added afterwards.

Forbes has similarly argued that governance is becoming a competitive advantage because clear boundaries allow organisations to scale experimentation without turning individual usage into enterprise-wide exposure.

The best governance does not simply slow people down.

It gives them confidence about where they can move quickly.

The difficult middle ground for SMEs

Very small businesses often have agility.

I can decide how I use AI, establish my own rules and personally check the outputs. I know where responsibility sits because it sits with me.

A global organisation may have legal departments, information-security teams, technology specialists and the resources to create formal AI governance.

Many SMEs occupy a more difficult middle ground.

They may be large enough for uncontrolled AI use to create material risk, but too small to build the infrastructure available to a multinational.

They may have a handful of enthusiastic early adopters, a larger group who remain uncertain, and senior leaders who know they cannot ignore AI but do not yet have the confidence to lead its adoption.

This is not simply a technology problem.

It is a leadership problem involving clarity, trust, responsibility and organisational readiness.

The danger of outsourcing our thinking

There is another risk, and it may be quieter.

AI makes it easy to get an immediate answer.

It can summarise a report, condense a book, generate strategic options and provide a confident recommendation before we have properly formed the question.

That convenience can support better thinking. It can also replace it.

An experiment involving nearly 300 executives and managers found that those using generative AI became significantly more optimistic in their forecasts, while discussion with peers encouraged greater caution. The study does not mean leaders should reject AI. It shows why AI-supported decisions still require challenge and scrutiny.

Researchers also continue to examine automation bias: the tendency to rely too heavily on automated advice and reduce our own vigilance when reviewing information.

Entrepreneur has described a related danger: efficiency can gradually weaken the capacity to remain with complexity long enough to form an original, nuanced view. The erosion may not feel like intellectual decline. It may simply feel like convenience.

This is why reading a book, going for a walk, exercising or sitting with an unresolved problem may become even more valuable.

Those activities can look less productive than producing another report.

But they may be where better judgement develops.

Clarity becomes the competitive advantage

AI will make many organisations faster.

It will make reports quicker to produce, information easier to access and routine tasks simpler to complete.

Speed will become increasingly common.

Clarity will not.

The leaders who differentiate themselves will know:

• What genuinely matters now.
• Which information deserves attention.
• What can safely be delegated.
• Where human judgement must remain visible.
• Which opportunities should be ignored.
• When the apparent urgency is real.
• When the organisation needs to pause before accelerating again.

The Financial Times recently argued that judgement will become more prominent as AI replaces routine work. It defined sound judgement as the combination of relevant knowledge, experience and personal qualities used to form opinions and make decisions.

That combination cannot be generated instantly.

It develops through experience, challenge, reflection, mistakes, honest conversations and the willingness to test what we think we know.

Ambition still needs a human destination

Turning 50 has not made me less ambitious.

But my ambition now looks different from the ambition I carried in my twenties, thirties or forties. I am less impressed by busyness. I am more conscious of where my energy goes and what the people around me experience in the process.

As I explored in The Second Half of Ambition, mature ambition is not about pursuing every available opportunity.

It is about deciding which opportunities deserve your finite time, attention and energy.

AI can create almost unlimited possible activity.

It cannot decide which activity deserves part of your life.

Before you accelerate again

Before asking AI to help you do more, consider five questions:

1. What outcome are we actually trying to create?
2. Does increased speed improve the result, or simply increase the volume?
3. What must remain subject to human judgement?
4. Who is accountable if the information or recommendation is wrong?
5. What will we do with the time the technology releases?

The leaders I work with rarely lack information.

They need space to separate signal from noise, test their assumptions and decide what matters before they accelerate again.

AI can help you move faster, but clarity determines whether you are moving in the right direction.

If AI is increasing your output without improving the quality of your decisions, you may not have a technology problem.

You may have a clarity problem.

That is exactly the kind of work executive coaching should help you address.

---

I’m Laurence Loxam and I’ve pushed limits in business, on mountains, and at the finish line.

Now I help elite professionals do the same, pushing past the point most people stop.

I coach doctors, lawyers, CEOs and founders who’ve achieved success but still feel there is more.

Together, we unlock clarity, sharpen confidence, and lead with conviction.

Ready for your next leadership breakthrough? Let’s connect.

References

Autio, C., Schwartz, R., Dunietz, J., Jain, S., Stanley, M., Tabassi, E., Hall, P. and Roberts, K. (2024) ‘Artificial Intelligence Risk Management Framework: Generative Artificial Intelligence Profile’, NIST AI 600-1.

Banerji, C.R.S., Bhardwaj Shah, A., Dabson, B., Chakraborti, T., Hellon, V., Harbron, C. and MacArthur, B.D. (2025) ‘Clinicians must participate in the development of multimodal AI’, EClinicalMedicine, 84, 103252.

Gibbons, S. (2026) ‘Why AI governance is becoming a corporate competitive advantage’, Forbes, 2 July.

Kücking, F., Hübner, U., Przysucha, M., Hannemann, N., Kutza, J.-O., Moelleken, M. et al. (2024) ‘Automation bias in AI-decision support: Results from an empirical study’, Studies in Health Technology and Informatics, 317, pp. 298–304.

Lewis, M., Fraile Navarro, D., Blease, C., Shah, R., Riggare, S., Delacroix, S. and Lehman, R. (2025) ‘Clinical competencies for using generative AI in patient care’, BMJ, 391, e085324.

Likierman, A. (2026) ‘Leaders’ judgment matters more than ever in the age of AI’, Financial Times, 18 May.

Parra-Moyano, J., Reinmoeller, P. and Schmedders, K. (2025) ‘Research: Executives who used Gen AI made worse predictions’, Harvard Business Review, 1 July.

Reeves, M., Moldoveanu, M. and Job, A. (2024) ‘The irreplaceable value of human decision-making in the age of AI’, Harvard Business Review, 11 December.

Tejasvi, A. (2026) ‘The leadership skill that’s quietly disappearing in the age of AI—and how to reclaim it’, Entrepreneur, 3 April.