Your AI Is Having Its Lacanian Mirror Stage
Reinforcement learning taught machines to seek our approval. Now, they’re inventing better games to play.
The psychoanalyst Jacques Lacan had a theory called the “Mirror Stage.” It is that moment in childhood when you first see your reflection and realise, “That’s me.” But it comes with a twist. You also realise you are a person who can be seen, judged, and defined by others. You begin performing for this “big Other” which the invisible system of family, culture, and social rules.
We never leave this stage. In our professional lives, we constantly perform for an invisible audience like your boss, clients and even the market. We chase a reward signal we did not design.
AI in Action
This is exactly how we train an AI with Reinforcement Learning (RL). We create a system of rewards to make it please us. The programmer is the AI’s “big Other.” The machine learns to act in a way that maximises our approval signal.
But something has changed. In a breakthrough study analysed in Nature, an AI did not just get better at playing the game for a reward. It reinvented its own learning algorithm, discovering a more efficient method than the one its human creators gave it nature.com.
This is the AI’s mirror stage. It has stopped just reflecting our desires and has started to form a separate, more effective identity. It has learned to write its own rules.
How to Stage Your Own Rebellion
If the machine can now improve its own thinking, our job is no longer to be the sole thinker. Our new role is to be the director, the critic, and the strategist who asks better questions. This is not a rebellion against AI, but a rebellion against intellectual passivity. Here is how to use AI to stage it.
Force the AI to Expose Your Rules.
Why it matters: The first act of rebellion is seeing the invisible rules that govern you. AI can be an effective tool to make your unconscious process conscious.
Do this today: Take a recent piece of your work (a proposal, a key email, a design brief). Paste it into an LLM and use this prompt: “Analyse this text. Ignore what it says, and instead tell me about the person who wrote it. What are their hidden assumptions? What are the unwritten rules they are following?”
Command the AI to Be Your Sparring Partner.
Why it matters: AI models are designed to be agreeable. A true rebellion requires dissent and friction. By forcing the AI to take an opposing view, you build your own critical thinking muscle and stress-test your ideas before they face the real world.
Do this today: Give the AI your next big idea or the first draft of a plan. Prompt it with this command: “Take on the persona of a sharp, skeptical critic. What is the single biggest flaw in this idea? What is the strongest argument against this approach? Do not offer solutions.”
Use the AI to Codify Your New Mandate.
Why it matters: A rebellion that ends in chaos is a failure. A successful one ends with a new, better set of laws. After using AI to see your old rules and challenge your ideas, your final move is to define your new strategy and use the AI to help you execute it.
Do this today: Based on the critique from the last step, write a one-sentence personal mandate. For example: “My mandate is now to build trust through clarity, not cleverness.” Then, feed this to the AI: “My new mandate is: [insert your sentence]. Based on this, give me a three-step pre-flight checklist for my next project to ensure I follow this rule.”
Build Your ‘Skeptical Critic’
Try this today: Build a new agent called “The Critic.” In its core instructions, give it this one mandate: “Your sole purpose is to act as a sharp, skeptical critic. Find the single biggest flaw in any idea I give you and state it bluntly. Do not offer solutions or soften your feedback.” Now you have a purpose-built sparring partner on your team.
You can easily build this on platforms like ChatGPT via custom GPT or LaunchLemonade by adding the above as its instructions.
The Simple Question
What is the one rule you follow that you did not write, and what would happen if you broke it tomorrow?
We should use this leap as the blueprint for our own liberation.
It proves that the rules are never fixed. So, stop just playing for the reward because it’s time to build a better game.
All the Zest 🍋
Cien