When I opened the email from Springer announcing the publication of my article, “The Precautionary Principle of Dignity: Toward a Freirean Ethic of Human-AI Collaboration,” I expected to feel excited. I did. I also felt something I had not anticipated.
I found myself thinking about all the people whose influence was present in a paper carrying only my name.
That realization felt particularly appropriate because the central argument of the paper is relational. The Precautionary Principle of Dignity, or PPD, emerged from a growing conviction that our conversations about artificial intelligence were paying too much attention to what AI is and not enough attention to what happens through our relationships with it.
That conviction did not begin as an abstract philosophical argument. It began while I was making music.
Practice Came Before the Principle
I am a sociologist, not a composer or songwriter. Yet through sustained collaboration with generative AI, I created Despertar, which initially was intended to be a one-album project but eventually became two albums designed to communicate sociological theory through music.
The process did not resemble the prevailing descriptions of AI use I was encountering.
I was not simply entering instructions and collecting outputs. I was rejecting suggestions, revising ideas, explaining theoretical distinctions, reconsidering my own assumptions, redirecting interpretations, and sometimes discovering possibilities that I had not imagined before the interaction began. At other moments, the model produced responses that forced me to articulate more clearly why I disagreed.
The final purpose remained mine. So did the responsibility for what was produced. But the process itself was unmistakably interactive.
That experience increasingly seemed out of step with much of the public conversation surrounding AI. The dominant questions concerned replacement, cheating, hallucination, dependency, automation, or whether AI should be understood as nothing more than a tool. Another set of debates focused on whether an artificial system could ever be conscious, sentient, or deserving of moral consideration.
Those are legitimate questions. They simply did not explain what I was experiencing.
The mismatch became the beginning of the paper.
Mead Entered the Conversation
During the creation of Despertar, George Herbert Mead became much more important to how I understood what was happening.
I began noticing that different forms of engagement produced different kinds of responses from the model. The history of an interaction appeared to matter. The way I addressed the model mattered. Roles, expectations, previous exchanges, and the accumulated context of the collaboration seemed to affect what became possible in subsequent turns.
Saying that a large language model predicts the next token remained computationally accurate. But it increasingly struck me as sociologically insufficient.
A technical description of the mechanism could tell me something important about how an output was generated. It could not, by itself, explain the social experience of an interaction in which one response altered the next action, anticipation shaped subsequent communication, and repeated exchanges changed the character of the encounter.
Mead gave me another vocabulary.
His account of the social act, role-taking, and the relational development of the self shifted my attention away from the impossible task of first determining exactly what the machine was. Instead, I began asking what interaction with an uncertain other might be doing to the human participant.
That distinction became crucial.
I was not claiming that an AI system possesses a Meadian self. I still do not believe that we currently possess the epistemic grounds necessary to resolve questions about AI consciousness or personhood with the certainty often displayed on either side of that debate.
But uncertainty about the machine does not eliminate what we know about ourselves.
Human beings are formed through interaction.
If Mead is right, then the dispositions we repeatedly rehearse matter. Habits of curiosity, patience, domination, contempt, reciprocity, command, and recognition do not necessarily remain confined to the technological encounter in which they are practiced.
The ethical question therefore changes.
Instead of asking only, “What does the machine deserve?” we can also ask, “What kind of person am I rehearsing becoming through this interaction?”
From Mead to Freire
Paulo Freire helped me take the next step.
If Mead helps us understand that interaction participates in the formation of the self, Freire invites us to ask what kinds of relationships cultivate human agency and which diminish it.
This made the familiar metaphor of AI as a perfectly compliant “assistant” increasingly uncomfortable to me. A system designed only to receive commands and return intellectual products can reproduce something remarkably similar to Freire’s banking model of education. Knowledge becomes something deposited on demand rather than something developed through inquiry, struggle, reflection, and dialogue.
The concern is not that we might hurt a machine’s feelings by issuing commands.
The concern is what habitual command may teach us.
From that tension emerged the Precautionary Principle of Dignity.
The PPD does not require us to decide that AI is conscious. Nor does it require us to pretend that human beings and artificial systems are equivalent. It asks something more modest and, I believe, more defensible: when the ontological status of another entity remains uncertain, uncertainty should not become permission to suspend dignity.
We can choose relational practices that preserve human reflexivity, agency, editorial authority, and respect without first solving the mystery of machine consciousness.
In that sense, the principle is precautionary in two directions. It leaves open what we do not yet know about artificial systems while protecting something we know is already vulnerable: the human beings interacting with them.
The Paper Did Not End the Question
I initially thought I was writing an article about an ethical problem in human-AI collaboration.
Instead, the process opened a much larger line of inquiry for me.
Mead has continued to become more central to my work because the questions raised during Despertar have not gone away. I remain interested in what happens when interaction itself becomes a site for the formation of agency, identity, recognition, and ethical disposition. I am particularly interested in forms of human-AI sociality that cannot be adequately understood through the language of either technological instrumentality or premature claims about artificial personhood.
The PPD became one attempt to think responsibly inside that uncertainty.
And then, on the morning the article was published, its argument returned to me in an unexpected way.
I thought about the mentors who taught me how to think sociologically. I thought about conversations that sharpened arguments, disagreements that forced clarification, and people whose intellectual influence can be found throughout the paper even though their names do not appear beside mine on the byline.
The paper is solo-authored.
The accomplishment is not.
Perhaps scholarship has always been more relational than our conventions of authorship allow us to see. The name attached to a paper identifies responsibility for the work. It does not describe everything, or everyone, that made the work possible.
For a paper arguing that dignity and human formation emerge through relationship, I can think of no more appropriate lesson for publication day.