Who Writes When We Write Together?

Monday, July 13, 2026

Who Writes When We Write Together?

A Practical-Philosophical Reflection on Artificial Intelligence, Research, and Responsibility

A recent commentary in Science discussed a researcher who had published a peer-reviewed scientific article written entirely with the assistance of artificial intelligence. The researcher made no attempt to conceal the process. He had developed the ideas, assembled the material, and guided the entire project, but he allowed artificial intelligence to produce the text. His explanation was both simple and striking: “The article has my soul and AI’s words.”

The accompanying commentary was critical. While the practice might be legally permissible, it was described as ethically problematic from the perspective of research integrity.

I understand the concern. Yet I believe the question is far more complex than a simple yes or no.

What, in fact, are we evaluating when we evaluate a scientific publication?

Is it the text itself?

Or is it the thinking from which the text emerges?

Throughout my academic life, I learned that research begins long before the first sentence is written. It begins with curiosity, questions, reading, doubt, conversations, lived experience, and the slow maturation of ideas. Writing is essential, but it is only one part of a much larger intellectual process.

Today, I work differently from the way I did when I wrote my doctoral dissertation in 2009.

Back then, I worked alone with books, notebooks, and a computer. Ideas developed in silence. Every paragraph was written, revised, discarded, and rewritten. The process could continue for months.

Today, much of that same process unfolds through dialogue with artificial intelligence.

I ask for counterarguments.

I ask for tighter formulations.

I reject entire drafts.

I rearrange sections.

I ask for new perspectives.

I discuss references.

Sometimes I disagree. At other times, I discover that the text expresses something I have been trying to articulate for years.

Yet one thing has not changed.

It is still I who decide what remains on the page.

This experience has led me to ask a different question from the one that usually dominates the debate.

Perhaps the crucial question is not:

Did artificial intelligence write the text?

Perhaps the more important question is:

Who did the thinking?

If artificial intelligence performs the intellectual work, draws the conclusions, and the researcher simply publishes the result without critical examination, it becomes difficult to see how scientific responsibility can be maintained.

But if the researcher develops the research question, evaluates the arguments, verifies the sources, rejects weak reasoning, and accepts full responsibility for the final manuscript, artificial intelligence assumes a different role.

It is no longer the researcher.

It becomes a tool.

Yet this is not the whole story.

Language is never merely a technical surface. Hans-Georg Gadamer reminds us that understanding takes place through language. Words do not simply express thought; they also help to shape it. When artificial intelligence suggests formulations, it may therefore reveal new relationships between ideas that the author had not previously recognised.

This is both the opportunity and the challenge.

The opportunity is that dialogue can deepen our thinking.

The challenge is that the author may stop thinking independently.

Research ethics, therefore, is not fundamentally about who composed the sentences. It is about who bears responsibility for the truth they express.

My own experience is that artificial intelligence has not replaced the writing process. It has transformed it.

It resembles not a secretary but an inexhaustible conversation partner. Such a dialogue may challenge my arguments, suggest a different structure, or remind me of a philosopher I had overlooked. But it cannot live my life, acquire my experiences, or assume my responsibility.

It is still I who must decide whether the text expresses something true.

For that reason, I do not believe that the central question of future research ethics will be whether a text was written with artificial intelligence.

The more important question will be whether the author can still publicly defend every argument, every interpretation, and every conclusion.

Perhaps that is the real test.

Not whether the machine can write.

But whether the human being is still willing to think.

If artificial intelligence becomes a substitute for judgment, research loses something of its soul.

But if it becomes a dialogue partner that sharpens our judgment, it may strengthen rather than weaken scholarly inquiry.

In the end, the value of research is not determined by the words themselves.

It is determined by the intellectual integrity of the person who signs their name beneath them.