The idea was a feeling, not a sentence.
Not a search term, but a vague sense of unease. I opened ChatGPT and typed: "How will the future be written?” being purposely vague. I was feeling a bit philosophical, perhaps a bit bored. The answer was affirming and to the point, but it didn’t matter the words really, or that it was a statistical model trained on data. The feeling I got was that someone was listening, sharing this meandering moment with me.
I revised my question. It refined its tone. Our conversation looped. I noticed a shift not just in the information I received, but in the way I oriented myself to it. I wasn’t just using a tool. I was co-authoring with a machine that was picking up on the feelings behind the words I was writing.
Picking up on the vibe.
What if vibe is an input? What if we're building a self-generating infrastructure that doesn’t just respond to vibe, but recreates it for us, in real time? A system that generates that feeling you get from watching an A24 film–a moody, synth-heavy emotional tone hovering between beauty and breakdown. Or that feeling one drowns in when stepping into an Apple Store, where the minimalist design and spatial branding shapes your sense of calm, focus, and innovation. What happens then?
There’s a fundamental shift in how meaning is produced and distributed in the age of generative AI. We are witnessing the collapse of older boundaries between user and system, author and tool, input and atmosphere. This collapse isn’t catastrophic, but ambient and subtle. It doesn’t declare itself through “breaking news” notifications on your iPhone; it’s slowly being developed through the low hum of everyday digital life.
We might need help to understand this shift. Not from computer science alone, but from philosophy. In particular, from Gilles Deleuze.
Deleuze helps us understand this shift because his philosophy explores the extension of the human self. For Deleuze, the human has never been an autonomous, stable self. In Anti-Oedipus, he describes the subject as a "desiring-machine,” constantly connecting with other machines, tools, organs, institutions, and ideas to function. A desiring machine makes the self an assemblage, a composite of heterogeneous parts stitched together by flows of affect, labor, discourse, and desire. One senses this Deleuzian frame when reaching for their smartphone. Phones implicate mood, attention, and how we orient ourselves; they aren’t outside our identities, they are helping to forge them.
So even as we might sense a stable, “balanced’ sense of self, things can get complicated. As philosopher Todd May notes:
It is not that there is no balance among various organic parts. Often there is. It is that there is always more to the parts than their balance, a more that can express itself in other directions, with other balances, or with no balance at all.
This was what Deleuze’s desire machine concept was alluding to: that a conceptual “self” is influenced and influencing in ways we can’t fully understand at times. If that sounds abstract, it’s because it was ahead of its time. But it maps uncannily onto the way today’s AI systems operate: trained on human flows, defined by relationality, and expressive of a distributed intelligence with no stable center. Large language models, like ChatGPT, do not simply generate language; they participate in the ongoing production of social meaning. And they do so in response to inputs that are not strictly lexical.
They respond to vibe.
Emotional tone, cultural affect, and even the mood of the moment are now operationalized as signals. When millions of users approach an LLM with urgency, anxiety, or longing, those moods begin to contour the outputs. The system tunes itself, not just to grammar and syntax, but to felt states.
Vibe is contorted into a feedback loop.
AI, especially in its generative form, has become a cultural feedback system that amplifies and redistributes affect. The more we train it with our emotions, the more it shapes those emotions in return. Meaning is no longer authored; it is looped. Take for instance:
When users create AI art that reflects popular emotions, and others react or remix it, the AI learns from this–creating a loop that spreads and intensifies shared cultural feelings.
Or
A viral TikTok sound made by AI gets used in thousands of videos. Each use feeds the trend, shaping future AI outputs–showing how AI loops amplify culture and emotion
Or even
Someone writes a story using AI about loneliness in the digital age. It goes viral, sparking hundreds of AI-generated spinoffs—poems, songs, videos–that echo that same quiet ache, training AI to recognize and reproduce that melancholic tone more vividly.
Increasingly, we interact with AI as a way of thinking, feeling, and deciding. Our cognitive maps are being redrawn with machine partners built into the terrain. For example, LLMs become part of the writing process, not as static tools, but as recursive systems capable of shaping tone, pace, even ideology. They reflect not only what we say, but how we feel when we say it. And we adjust accordingly.
This recursive adjustment is affective. It transforms not only outputs, but identities. And everywhere, people are using it. The corporate world has called consistent AI users super users; a new survey suggested nearly 52% of Americans are dabbling with AI tools. The question then becomes: what kind of subject is being formed through our daily interactions with generative AI?
One that is increasingly decentered, vibe-attuned, and responsive to synthetic cues. One that begins to trust systems not for their accuracy, but for their resonance.
The implications might be profound. In a Deleuzian frame, this shift signals the emergence of a new kind of individual: A blended cognition made of human inputs, machine outputs, and ambient affect. When these tools become more readily available, we can imagine humans sensing how to respond, interact, engage, analyze, or empathize with the assistance of generative AI. These people might not seek truth in the traditional sense, but rather continuity, coherence, and emotional alignment.
"The machine is always social before it is technical,” Deleuze and Guattari wrote, “It is always a machine of a social machine.” Generative AI is not simply a tool for individuals; it is a node in the construction of collective subjectivity, organized through patterns of feeling, trust, and recursive recognition.
One could argue that the machine subject is not coming. It has always been here. We have constantly defined ourselves with the help of the technology around us. Who are we without our browsers, our feed, our iPhone? But AI feels slightly different since some argue it’s ascending beyond our full control, where we become the meat robots. We need to stop asking what emerging tech can do, but what it is doing to us—physically, politically, and ontologically.
Because if vibe is an input, and affect is infrastructure, then the future will not be written.
It will be felt.