AI and the Illusion of Mastery
I have been thinking about a passage Rachel Pollack wrote years before any of us were discussing ChatGPT, generative AI, large language models, or whether a machine has any business helping someone write a sentence.In “Pandora’s Bust,” Rachel describes a conflict between a “machine-centered world” and a “human-centered world.” At first, the distinction seems familiar enough: human against machine, agency against automation, the living imagination against the systems we have built. Rachel, however, makes the problem considerably harder. Her character, Michelangelo, believes he understands the conflict and, more importantly, believes he can master it. He imagines himself as the great humanist, capable of manipulating machines without being changed by them. Then comes the recognition: “But the machines did a better job of using me.”
I keep returning to that recognition, particularly now, when the question of our relationship with machines has moved beyond speculation and into the ordinary routines of our lives. Decades after Rachel wrote that scene, I brought it directly to an artificial intelligence. I did so deliberately, driven by a curiosity about what might happen if I asked the machine to look at itself. I gave it Rachel’s words and posed the question that had begun to emerge for me from them: What is the machine teaching us to become? There is an obvious irony in using a machine to interrogate our relationship with machines, yet I do not think that irony invalidates the inquiry. If anything, entering the exchange rather than observing it safely from outside may be where the more difficult questions begin.
Our public discourse about AI has increasingly arranged itself around rigid oppositions: embrace it or resist it; it will liberate human creativity or ultimately destroy it; it will democratize knowledge or flood the world with synthetic refuse. We name it according to the position we have already taken: tool, thief, assistant, threat, inevitability. Beneath that rhetoric are genuine concerns. Writers and artists have serious questions about how their work has been used, credited, and compensated. Workers have reason to wonder what automation will mean for their livelihoods. Questions of misinformation, environmental cost, concentrated corporate power, and our growing willingness to trust systems whose answers can sound authoritative even when they are wrong also loom.
Rachel’s story, however, keeps drawing me towards a question that sits beneath those arguments. What happens to us while we are using the machine? What habits of thought and attention are being altered so gradually that we may fail to recognize the change as it happens?
The machine may be teaching us impatience. Why struggle with a sentence for twenty minutes when something can produce twelve versions in seconds? Why read the book when we can ask for its argument? How long do we remain confused before we turn towards inquiry? Perhaps more consequentially, it may be teaching us to become less tolerant of uncertainty. Ask a machine a question, and it will generally offer an answer. Human thought is less accommodating. Some questions deserve to remain unanswered for a time. Some require us to set them aside, returning to them only after we've gained experience and insight. Rachel understood this remarkably well. Throughout her work, she allows uncertainty to remain, opening space for imagination and further thought.
Smaller exchanges, too, are so ordinary they can pass almost unnoticed. A navigation system calculates the fastest route and redirects us around an accident. Medical technologies help physicians detect patterns in scans, monitor patients, and identify possibilities that might otherwise be missed. A refrigerator can keep track of what is inside and suggest what we might make for dinner. We ask machines to remember this for us, summarize that, compare our choices, write the response, or interpret what happened. We may even ask them to tell us what we think.
These exchanges are clearly not equivalent, nor would I want to suggest that they are. Some extend human capacity in extraordinary ways; others relieve us of mundane tasks and leave our attention available for something else. Yet together, they make the boundary between assistance and surrender harder to locate. When does a technology extend our ability to navigate the world, and when does it begin to change the way we navigate it? When does assistance deepen human judgment, and when does it quietly replace the effort through which judgment is formed? These questions lead me back towards Michelangelo and the unsettling reversal at the center of Rachel’s story: at what point does using the machine become allowing the machine to establish the habits through which we encounter the world?
I do not think the answer requires pretending that we can stand outside technology. I used AI as part of the backbone of my investigation in writing this essay, deliberately placing the machine inside the question I was trying to understand. More generally, it has become one tool among many in the way I research and write. I still work longhand in notebooks with a fountain pen, sit with books and annotate their margins, spend time in libraries and archives, both online and in person, and follow questions whose answers may take weeks or months. I talk with other writers, revise, reconsider, and change my mind. I may ask AI to show me something I have overlooked, sort a complicated sequence of events chronologically, or point out a word I have repeated so often that I can no longer see it. At other times, I reject what it offers or close it entirely because the work requires something it cannot do for me.
Working this way has made me more interested in where discernment enters the exchange. When I read something AI has produced and recognize that it is vague, bloodless, too easy, or simply does not sound human, I have to ask what I am recognizing as human in the first place, and how I know when it is absent. I still have to decide what is useful, what deserves further investigation, what needs to be checked against a source, and what to discard. Perhaps the greater danger lies in reaching the point where we no longer notice the difference between being assisted in our thinking and having the thinking done for us.
Near the end of the passage in “Pandora’s Bust,” Michelangelo remembers using a telephone to communicate with another machine. Suddenly he sees himself differently: “I’d become a bastard machine myself.” Rachel’s reversal is striking. The machine does not become human; the human becomes machine-like. Michelangelo’s realization pushes the inquiry beyond what artificial intelligence can do and towards what we may gradually cease doing for ourselves.
What qualities are we unwilling to surrender for convenience? Attention, discernment, curiosity, doubt, memory, difficulty, imagination, relationship, and the capacity to remain with something we do not understand without immediately demanding an answer all seem worth guarding. Yet Rachel gives us another word, one I might not have placed on that list without her: grace.
Grace resists technological measurement. We cannot easily quantify it through speed, efficiency, productivity, accuracy, or output. It requires humility about mastery itself and an awareness of relationship, consequence, and limits. Michelangelo believed he could use the machines without being altered by them. His failure was not engaging with the machine. He believed the exchange moved in only one direction.
Perhaps this is what Rachel’s story offers us in the present moment. The question is larger than whether artificial intelligence is good or bad, whether we should embrace it or fear it, or even where we should draw the boundaries around its use. We are already living inside the exchange. The more difficult task may be learning to recognize what is passing between us and the machines we have made, what we are gaining, what we are surrendering, and what we may be becoming before the change becomes too familiar to see.
I asked the machine what it was teaching us to become. I am beginning to suspect that the answer will depend, in no small part, on what we refuse to stop being.
About the Story
Rachel Pollack’s “Pandora’s Bust” first appeared in New Worlds Quarterly 2 (1971), edited by Michael Moorcock. A surreal satire in which technology, government, religion, sexuality, and human ambition collide, the story raises questions about control, agency, and our persistent belief that we can remain somehow separate from the systems we create.











