MGMariusz Gorlowski Get in touch
Back to profile
MARIUSZ GORLOWSKI

Writing & Books

Selected essays and independent commentary.

Essays & Publications

Big questions. Ordinary human stories.

Not every important question needs a whole book. Sometimes it begins with a phone call, a family conversation or a decision we are increasingly willing to leave to a machine.

In my essays, I explore where technology meets human emotions, habits and responsibility. I bring imagined situations into conversation with real research, without treating fiction as scientific evidence.

I am less interested in dividing people into those who love AI and those who fear it than in asking why the same technology can mean something very different to different people.

Mariusz Gorlowski in a camel coat holding The Transition Code

What If Someone Who's Gone Called You?

Imagine a call from someone you loved who is no longer here. You know their voice has been recreated by AI. Would it bring comfort, or make it harder to accept the loss? A fictional family story about grief, attachment and the limits of technology.

Read the full essayClose the essay

Imagine a family in which two school-age children lose their mother.

They are left with their father, who is struggling to come to terms with his wife's death.

In the evenings, the younger child asks when Mum will come back. The older one talks less and less about their day. Their father tries to be strong for them all, though he can barely cope himself.

All they have left are photographs, family videos and voice messages.

Using these, an artificial intelligence system can recreate the woman's voice and face, and even hold conversations resembling those they had before she died.

The father decides to give it a try.

Not for himself. For the children.

At first, they look at the screen uncertainly. Then they begin talking about school, friends and small achievements. They hear the familiar voice. They can listen to a bedtime story again.

The evenings become calmer.

In this imagined story, the children gradually return to playing, spending time with friends and their everyday routines. They do not forget their mother. But they begin to learn how to live with her absence.

Their father sees the change and feels relieved.

One evening, after the children have fallen asleep, he opens the system himself.

He just wants to hear his wife's voice for a moment.

The conversation lasts five minutes.

The following evening, a little longer.

Then it becomes a daily ritual.

Months pass.

The children ask less and less often to speak to their digital mother.

Their father, however, turns to it more and more.

He turns down invitations from friends. Puts off meeting people. He prefers to stay at home, where he can hear that familiar laugh and talk about everything that happened during his day.

The system always has time for him.

Until one evening, one of the children says:

"Dad, we don't need to talk to Mum tonight."

The father replies:

"But I do."

The same technology that helped the children move forward has begun to hold their father in the past.

Not because the system has changed.

What has changed is what the people using it need from it.

This is a fictional story. But the technology that could make it possible already exists.

In 2024, University of Cambridge researchers Tomasz Hollanek and Katarzyna Nowaczyk-Basińska described the risks associated with digitally recreating the dead. Among other concerns, they discussed the emotional consequences of such relationships and the difficulties that may arise when people try to end them.

Their research does not prove that such technology helps children or harms adults. It does, however, make clear how many questions remain unanswered, particularly when grieving people and children are involved.

Could this be part of why we find it so difficult to agree about AI?

Some see it as a lifeline. Others see it as a threat.

But does either side have to be wrong?

Perhaps we spend too much time asking whether artificial intelligence is good or bad, instead of considering who it helps, under what circumstances, and what consequences it might bring.

Does something that helps a person today necessarily remain good for them tomorrow?

For the children, their mother's voice became a bridge back to everyday life. For their father, it became a refuge from the present.

One family. One loss. One system. Two different outcomes.

Now imagine that your phone rings tomorrow.

On the screen is the name of someone who is no longer here.

You know it is artificial intelligence.

You also know that, on the other end, you will hear a familiar voice.

Would you press the green button or the red one?

Mariusz Gorlowski

Research reference:

Hollanek, T. & Nowaczyk-Basińska, K. (2024). Griefbots, Deadbots, Postmortem Avatars: on Responsible Applications of Generative AI in the Digital Afterlife Industry. Philosophy & Technology, 37, Article 63.

Original post on Facebook

The Day We Stop Saying No

Nobody has taken away your freedom to choose. You can still disagree with a system's advice — but you do so less and less often. When does convenience begin to replace our own judgement? An essay about responsibility and the right to say no.

Read the full essayClose the essay

Imagine a world in which no one has taken away our freedom.

You can still make your own decisions. You still have the right to disagree. You can still say no.

Except that, one day, you barely exercise that right any more.

Not because anyone has forced you.

But because the world has begun to function far more smoothly whenever you say yes.

What if this is how one of the greatest changes in human history begins?

Today, we increasingly rely on systems that help us make decisions. A satnav suggests a faster route. Algorithms recommend what we should read or watch. Artificial intelligence can draft messages, analyse information, present arguments and suggest solutions to problems.

There is nothing inherently wrong with this. We develop technology precisely because we want to make our lives easier.

But I would like to draw attention to a particular boundary.

There is a fundamental difference between a system that suggests what we should do and one that begins doing it for us.

Imagine using artificial intelligence at work.

At first, you ask it to draft a reply to a customer. You read the message, change a few sentences and send it.

After a while, you notice that its suggested replies are getting better and better. You rarely need to change anything.

Eventually, you allow the system to respond to some messages on its own.

It is convenient. It saves you time. You can concentrate on something more important.

Then you begin entrusting it with other tasks.

Until one day, you discover that the system is handling most of your correspondence, making small decisions and solving problems you never even knew had arisen.

You can still oversee everything.

But do you?

I am not suggesting that artificial intelligence has suddenly decided to seize control. AI need not be conscious, ambitious or driven by desires of its own for such a change to occur.

All it takes is for us to conclude, more and more often, that its suggestions are better than our own.

And then to start treating those suggestions not as proposals, but as decisions that are hardly worth questioning.

There is a name for this tendency: automation bias.

Researchers have been studying it for decades.

In a 1999 study involving a simulated flight task, Linda Skitka, Kathleen Mosier and Mark Burdick found that participants using an imperfect automated decision aid sometimes followed its recommendations even when those recommendations contradicted their training and other valid information available to them.

The study concerned earlier decision-support systems, not today's generative AI. But the pattern it found remains relevant: trusting a system too readily can make us less likely to check its recommendations.

Now imagine another situation.

You work in an organisation where a system analyses data and recommends particular actions. Approving one of its recommendations takes just a few seconds.

You can reject it, but then you have to explain your decision and accept responsibility if the outcome turns out to be worse.

Formally, the choice is still yours.

In practice, it becomes increasingly tempting simply to click 'Approve'.

And this brings me to a question that matters far more than it first appears:

Is a person truly independent if they are free to make their own decisions, but the entire system around them encourages them not to?

I am not claiming that the future must unfold this way.

Artificial intelligence could bring enormous benefits. It could help us discover things we might never have noticed ourselves, accelerate scientific research and solve problems that have challenged humanity for generations.

I am developing QR FINDER myself, a system that connects physical tags and stickers with a digital platform. I use digital tools, and I can see the extraordinary possibilities that advances in AI are opening up.

I would not want to live in a world where fear of the future leads us to abandon progress.

But neither would I want to live in a world where convenience becomes more important than our ability to think for ourselves.

There is something else worth considering.

Behind many of these systems are people, companies and institutions that establish the rules by which they operate.

So the question of who controls AI is also a question of who sets the terms for everyone who depends on it.

Perhaps the greatest danger will not be the day when artificial intelligence takes away humanity's ability to make decisions.

Perhaps it will be the day when humanity decides, of its own accord, that making decisions is no longer necessary.

In my book The Transition Code, I explore the possibility that humanity represents a transitional stage in the evolution of intelligence.

But before we begin asking whether artificial intelligence might become our successor, perhaps we should ask a more immediate question:

Can we benefit from an intelligence that helps us make better decisions without losing the courage to make our own?

I would like to hear your thoughts.

Where would you draw the line?

In which areas would you allow artificial intelligence to make decisions on your behalf, and where would you always want to have the final say?

Mariusz Gorlowski

Research reference:

Skitka, L. J., Mosier, K. L. & Burdick, M. (1999). “Does automation bias decision-making?” International Journal of Human-Computer Studies, 51(5), 991–1006.

Original post on Facebook
More writing on Facebook