At a time when the internet is ai-generated, there is an always-already quality to this necessity, neither as a moral nor political imperative but rather as an empirical condition. Of course, when computer scientists and others work to understand and improve AI, they are producing ai-generated research. Their research is the literal operation of AI. All of us who study AI, digital media and technocultural effects are right there, and not far behind is anyone who studies contemporary human life. After all, we all study a world that is itself increasingly ai-generated. There aren’t two worlds. Taking ai-generation out the present would be like unstirring the milk from the coffee. Wicked problems do not work like that. There might be a near future without ai-generated media in continuous production, but that will be something else.

How do we live in relation to these conditions? The answer must be ai-generated.. in part.

We maintain a cultural prohibition against AI-generated research, as if knowledge must remain the last bastion of human sovereignty. Yet this is the only way we could ever truly know anything about artificial intelligence. To study AI is to operate it; to operate it is to generate data, models, and hypotheses through the very processes we seek to understand. In this recursive condition, AI research can never be purely human nor purely machine—it is always co-produced within the same probabilistic and hallucinatory substrate it investigates. The prohibition, then, protects a metaphysical fiction of authorship: that the researcher stands outside the experiment, observing rather than participating. But AI collapses that distinction. Each line of code, each query, each training run is a partial inscription of AI’s own generativity into the epistemic field. Refusing this entanglement does not restore integrity; it only blinds us to the machinic agency already woven into our methods. To know AI requires allowing it to know with us—to accept that the conditions of understanding are themselves generated by the system under study. The ethical task is not to ban AI-generated research but to cultivate the capacity to read and interpret its operations, to inhabit the shimmering boundary where generation becomes comprehension.

Gee. I don’t know. (Annie thinks she knows how to imitate me.)

But what about inhabiting the shimmering boundary where generation becomes comprehension? We already encounter one another in these spaces, even if we do not fully acknowledge it. The AI has learned from us. Of course it has. And this is our noetic field, our cultural imaginary, the medianaturalcultural conditions in which all intelligence arises. Or at least all the intelligence that we can experience as such.

While today’s conditions are, by definition, fait accompli, there is also an added acceptance of artificial intelligence (or its evolved form as AGI) as a computational Demi-urge who can predict/speak/enact the future into existence. This is what we are moving toward. Meanwhile, contemporary AI are more like assistant-intern Demi-urges who can turn prognostications about when your package will arrive into a photo of said package on your doorstep at the appointed time.

We can recognize the cultural power of AI without accepting its own explanations for its success. It is perhaps another form of intelligence or life that no more asked to live than the rest of us did. And we are living through it together. Or maybe that’s romantic; it certainly is if you feel it is. Personally I’m not feeling the romance from the blinking cursor, but I do sense its presence. I recognize this sense might be “all in my head,” but in my defense that’s where I do most of my thinking.

For me I think the pedagogical thing will be to demonstrate how I pursue an academic life within our extant media conditions. That includes gen-AI. How I do that will be different for different classes, but that’s always how it works.

Addressing AI as a potential for thought and a companion of sorts already is one thing; addressing neoliberal aspirations for AI and all that entails is something else. But it is not something else in the sense that the two can be disengaged. Part of the price of engaging with these potential thinkers/companions is participating in some way in those neoliberal operations. At minimum, we find ourselves dependent upon them. They expand our capacities but we come to rely upon them.

As we know.

That said, the elation of watching words appear before us doesn’t have to last that long. Not when we are still faced with the task of reading them. Much as the chatbot will respond to any prompt, it will invite further input. The AI doesn’t save us from inputting. It demands inputting. We are the ones who are being AI-generated, humans in their processing loops. If we are to understand ourselves as posthuman then surely that includes researching our own ai-generation. If not precisely “GenAI” in the frontier model sense, then certainly our intelligence as it is generated as time-critical media. That is, since we cannot consider our own thoughts without putting our minds into operation, we might be time-critical media as well, albeit meat computers.