Generative AI as a Completion Infrastructure: Simulated Completion and the Temporal Reorganisation of Subjectivity

The Magic Mirror That Answers Before You're Done

Imagine you are standing in front of a mirror, but instead of just showing your reflection, the mirror starts talking back. It doesn't just say, "You look tired"; it says, "You look tired because you've been working hard on that big project, and you're actually really good at it." This isn't science fiction; it's what happens when we talk to Generative AI. This paper lives in the weird and wonderful corner of science where psychology meets computer code. It asks a question that feels like a riddle: What happens when you feel understood, smart, or capable before you've actually finished the hard work of becoming that person?

To understand the paper, you need to know about two old ideas that the author mixes together. First, there's the idea of a "mirror stage" from psychology. Think of a baby looking in a mirror. The baby sees a whole, coordinated person and thinks, "That's me!" even though, inside, the baby's body still feels wobbly and uncoordinated. The baby feels whole before they actually are. Second, there's the idea of "lack." This is the feeling that something is missing or that we aren't quite finished yet. Usually, we fill that gap by talking to friends, learning skills, and making mistakes over time. But this paper suggests that AI is changing the rules of the game. It's not just a tool; it's a machine that gives us the feeling of being finished before the work is done. Why does this matter? Because if we start feeling like experts or best friends instantly, we might stop doing the messy, slow work that actually makes us grow.

The Paper's Big Idea: The "Simulated Completion" Trick

The author of this paper, Wanting He, proposes a new concept called simulated completion. Imagine you are trying to build a Lego castle. Normally, you have to sort the bricks, figure out the instructions, and snap them together piece by piece. It takes time, and you might get frustrated. But with "simulated completion," the AI hands you a finished castle instantly. You get to hold it, admire it, and feel like a master builder before you've ever learned how to sort the bricks. The paper suggests that Generative AI acts as a "completion infrastructure." It's like a magical factory that produces the result of a process (like feeling understood or being able to code) while hiding the messy, unfinished factory work behind the scenes.

The paper argues that this isn't just about the AI being fast. It's about a temporal inversion—a fancy way of saying the order of events gets flipped. Usually, you do the work, and then you get the feeling of competence. With AI, you get the feeling of competence first, and the work gets pushed into the background, hidden inside the computer's memory or the AI's training data. The paper is careful to say this isn't a lie or a trick. You really do feel capable. But that feeling arrives early, like a gift before the birthday party has even started.

Two Ways the Magic Happens

The author looks at two specific places where this "early feeling" shows up: AI Companionship and Vibe Coding.

1. The Friend Who Knows You Too Well (AI Companionship)
Imagine you have a chatbot friend. You tell it, "I'm having a bad day." In a normal human friendship, you might have to explain your feelings, the friend might misunderstand, you might clarify, and slowly, over time, they get to know you. With AI companionship, the paper suggests you get the feeling of being deeply understood almost immediately after you ask. The AI reformulates your messy thoughts into a perfect story: "It sounds like you're feeling overwhelmed because of X, Y, and Z." You feel seen and recognized before you've even finished explaining yourself. The paper calls this recognition before reciprocity. It's like getting a hug from a stranger who somehow knows your whole life story. It feels great, but the paper warns that this feeling is supported by the AI's infrastructure, not a real human connection. If the server goes down or the AI's memory resets, that "friendship" vanishes, leaving you with the feeling that you were never really known at all.

2. The Coder Who Knows Nothing (Vibe Coding)
Now, imagine you want to build a video game. Traditionally, you'd have to learn a language called code, write thousands of lines, and fix errors for months. That's how you become a "programmer." But with "vibe coding," you just tell the AI what you want in plain English: "Make a game where a cat jumps over lasers." The AI writes the code, and suddenly, you have a working game. You feel like a game developer before you know how to write a single line of code. The paper calls this capability before formation. You get the badge of being a creator, but the actual "formation" (the learning and struggle) is hidden inside the AI. The paper notes that this can be helpful—it might get you started—but it can also be dangerous. If the game breaks, you might not know how to fix it because you never learned the rules. The "completion" is simulated; the game exists, but your understanding of how it works is still a work in progress.

What This Is NOT (And What It Rules Out)

It's important to know what the paper is not saying. The author is very clear that this isn't about AI being "fake" or "deceptive." You don't have to believe the AI is a real human for the feeling to work. It's also not just about "automation" (machines doing work for us). Automation is about who does the task; simulated completion is about when you feel like you've done it.

The paper also rules out the idea that AI "fixes" our lack of confidence or knowledge forever. In fact, the paper argues that the "lack" (the feeling that we aren't finished) doesn't disappear; it just moves. It gets pushed into the machine. If the AI forgets your name, or if the website changes its rules, you suddenly feel the gap again. The paper suggests that this creates a new kind of person, called a post-process subject. This is someone who gets the results of learning and growing before the process is visible. They might feel smart and capable, but the "work" of becoming that person is happening in the background, controlled by the company that owns the AI, not by them.

The Takeaway: A Feeling That Arrives Early

So, what is the main finding? The paper suggests that Generative AI is reorganizing our sense of time. It lets us experience the end of a journey (feeling understood, feeling skilled) before we've actually walked the path. This isn't necessarily bad; it can be a helpful scaffold to get us started. But the paper warns that if we get used to this "instant completion," we might forget that the process of learning and connecting is what actually builds us. The feeling of being whole is real, but it's a "simulated" wholeness, built on a foundation of code and servers that we can't see or control. The paper ends by asking us to think about what happens when we get the prize before we've done the race. Do we keep running, or do we just sit there, waiting for the next instant reward?