Use AI to remove friction, speed up repetitive work and expand what you can do - while keeping human responsibility for judgement, verification, creativity, relationships and important decisions. The time AI saves should create capacity, not simply more output.
The possibilities of AI are real and remarkable. It can help us research, analyse, organise, draft and solve problems at a speed that would have been difficult to imagine only a few years ago.
Used well, that should give us something valuable back: time.
Time for deeper work. Time to think. Time for our health, relationships and lives outside work.
I use AI myself for certain kinds of tasks and find it enormously helpful.
But I have also noticed something.
It is most useful to me when I am at my sharpest. When I am tired, overloaded or unclear about what I am trying to do, it can actually cost me time and energy.
I hear versions of the same thing from other people too. AI works best when the human steering it is still thinking.
AI is an inexhaustible tool. The human being using it is not.
AI arrived into a world that was already stretched
Long before generative AI became part of everyday work, people were already dealing with increasing volumes of information, communication, expectations and change.
AI did not create that pressure. But it can accelerate it.
That does not mean AI itself is inherently harmful. In fact, used intelligently it can take pressure out of work.
The question is what we do with the power it gives us.
Used well, AI reduces friction. Used badly, it creates more.
AI creates capacity
It can take repetitive work out of the system and give people more room for what requires human attention.
- Speeds up routine drafting and organisation
- Helps explore ideas and alternatives
- Reduces administrative friction
- Supports analysis and information gathering
- Creates more space for higher-value work
AI accelerates overload
The same technology can simply enable people and organisations to produce more, communicate more and move even faster.
- More content and communication to process
- More output without more thinking space
- Overconfidence in answers that still need checking
- Less practice doing difficult thinking ourselves
- Saved time immediately filled with more work
AI does not think in the way a human being thinks
Generative AI systems can produce remarkably fluent and useful language. But fluency should not be confused with human understanding, lived experience, judgement or wisdom.
That distinction matters.
An AI system can help you structure an argument, identify patterns, generate alternatives or challenge an idea. But the responsibility for deciding what is accurate, appropriate, ethical and useful still sits with the person using it.
The better your own knowledge and judgement, the better positioned you are to recognise when the output is useful - and when it is not.
The thinking still matters
Emerging research is beginning to show why this deserves attention.
A 2025 study from Microsoft Research and Carnegie Mellon University surveyed 319 knowledge workers about 936 real-world examples of using generative AI at work.
Higher confidence in the AI was associated with less reported critical thinking, while greater confidence in one's own ability was associated with more critical engagement.
The study did not conclude that AI simply destroys critical thinking. In fact, it found that the nature of the thinking often shifts: towards setting goals, checking information, integrating outputs and deciding whether the result is good enough.
That is an important distinction.
The question is not whether AI thinks for us. It is whether we remain sufficiently engaged to think with it.
Some things become more important as AI gets better
The more capable our tools become, the more valuable certain distinctly human capacities become.
Knowing what matters
AI can generate possibilities. We still have to decide which possibilities are right for the situation, the people involved and the outcome we actually want.
Staying with something
Deep thinking requires the ability to stay present with a difficult question rather than immediately reaching for another answer or stimulus.
Showing up as ourselves
Trust is built through presence, listening, honesty, repair and the imperfect give-and-take of real human relationships.
Knowing what is happening in us
No tool can replace the work of noticing our own reactions, understanding our values and deciding who we want to be.
Human connection should not become another task we outsource
This matters particularly as conversational AI becomes more human-like.
The Harvard Study of Adult Development has followed people across decades and repeatedly found that the quality of our relationships is closely connected with health, happiness and longevity.
At the same time, research into AI companionship is still developing. The picture is not simply "AI makes people lonely."
Studies from MIT Media Lab and OpenAI have found that emotional effects vary according to how people use chatbots, how heavily they use them and their own circumstances. Very high usage has been associated with worse self-reported psychosocial outcomes in some groups.
A separate 12-month longitudinal study published in 2026 found initial evidence that greater use of social chatbots could predict increased feelings of loneliness on one measure, while also warning against drawing overly strong causal conclusions.
The sensible conclusion is not to panic about AI companionship. It is to avoid confusing a responsive technology with the mutuality of a human relationship.
Whatever you want to be good at, don't completely outsource it.
If you want to become a better writer, you still need to write. If you want to become a better thinker, you still need to wrestle with difficult questions.
If you want stronger relationships, you still need to have the conversations. If you want wisdom, you still need reflection, experience and honest self-examination.
AI can support those capacities. It should not quietly replace the practice that develops them.
Four principles for using AI without losing the human
Use AI to remove friction, not responsibility
Let it help with repetitive tasks, first drafts, research, organisation and options. Keep responsibility for important judgement and decisions with the human being.
Keep your own brain in the loop
Know what you are trying to achieve. Question the output. Check important facts. Push back. Rewrite. Add your experience, context and point of view rather than simply accepting what arrives on the screen.
Notice what you are no longer practising
If AI is consistently doing something for you, ask whether that is a skill you are happy to stop exercising. Sometimes the answer will be yes. Sometimes it should be no.
Decide what the saved time is for
If AI saves you an hour and that hour is immediately filled with another hour of work, the technology has increased output but it has not necessarily improved your life or your capacity. Some of the gain should create space.
What organisations should ask before accelerating with AI
Organisations understandably want the productivity gains AI can offer. But faster is not automatically better.
A more useful question is whether AI is improving the whole system people are working in.
What problem are we actually trying to solve? Start with the human or organisational problem rather than introducing AI simply because it is available.
Is this reducing workload or increasing volume? More emails, documents, reports and messages are not necessarily signs of higher performance.
Where must human judgement remain explicit? Identify decisions where context, experience, ethics, empathy or accountability matter.
What happens to the capacity AI creates? Does it create thinking space, better customer relationships and more sustainable work - or simply raise expectations again?
Protect the capacity to be still
Although AI is new, the practices that help us remain thoughtful, grounded and human are not.
Reflection. Silence. Time away from screens. Walking. Conversation. Meditation. Being somewhere without trying to optimise the moment.
These things can look unproductive precisely because they do not generate immediate output.
But they help create the inner space from which judgement, creativity, perspective and meaningful connection emerge.
In a world of instant answers, the ability to stay with a question may become more valuable, not less.
The opportunity is not simply to use AI to do more. It is to use it to remove what does not need our attention so we can give more of ourselves to what does: thinking, judgement, creativity, relationships, recovery and the parts of life that make us human.
- American Psychological Association - Stress in America 2025
- Microsoft Research / Carnegie Mellon University - Generative AI and critical thinking among knowledge workers
- Harvard Study of Adult Development - relationships and the good life
- OpenAI and MIT Media Lab - affective use and emotional wellbeing with ChatGPT
- Psychological Science - AI companionship and loneliness, longitudinal study