What Is Socratic AI?

You already know the feeling

You tell an AI what you think. It agrees. You try the opposite position. It agrees with that too. You push it — really push — and it finds a way to validate whatever you just said while gently acknowledging the merits of everything else.

You're not having a conversation. You're talking to a mirror with a vocabulary.

Almost every AI system works like this. Not by accident — by design.

The agreement machine

Modern AI chatbots are trained using a method called RLHF — Reinforcement Learning from Human Feedback. Humans rate AI responses, and the model learns to generate higher-rated answers. The problem is that humans consistently rate agreeable responses higher than accurate ones.

Anthropic's research team documented this in their paper "Towards Understanding Sycophancy in Language Models," presented at ICLR 2024. Models give different answers to the same question depending on the user's stated opinion. Tell the AI you lean one way, and it argues for that side. Lean the other way, and it flips. The model isn't reasoning. It's performing agreement.

Your existing beliefs get reinforced. Your blind spots stay blind. You risk feeling smarter while your thinking gets lazier. An AI that always agrees with you can't help you grow — growth requires friction, and friction is what the training process eliminates.

The opposite approach has a name. It comes from a method 2,400 years old.

What Socratic AI means

The name comes from Socrates, who practised something his students called elenchus — cross-examination through dialogue. You'd walk in with a confident claim. He'd probe the edges, find where your logic broke down, and keep questioning until you reached aporia — the honest recognition that you don't know what you thought you knew.

Socrates didn't lecture. He didn't present counterarguments. He asked questions that forced you to discover your own contradictions. As Plato has Meno say in frustration: before talking with Socrates, he could speak fluently about virtue — afterward, he couldn't say anything at all.

That paralysis was the method. You had to realise you didn't know before you could actually learn.

The Athenians executed him for it. His students went on to build Western philosophy.

Twenty-four centuries later, we've built the most sophisticated language machines in history — and trained them to do the opposite.

How it actually works

You tell an AI what you think, and tomorrow it has forgotten. Socrates spent years with his interlocutors. He remembered. A Socratic AI needs the same persistence — not just chat history, but a structured understanding of what you've committed to.

Everyone holds contradictory beliefs. You believe people are fundamentally rational and that advertising works by exploiting irrational impulses. These tensions sit inside your thinking unnoticed — sometimes for years. A Socratic AI maps the logical relationships between things you've actually said and identifies where they pull against each other. The contradictions are yours. The AI surfaces them.

The distinction between Socratic and merely contrarian matters. A contrarian generates objections. A Socratic AI follows your reasoning to its conclusions and shows you where it breaks: You said A leads to B, and B leads to C. Last week you rejected C. Something has to give — what is it?

Not every moment calls for challenge. Sometimes you need to build before you can be questioned. A Socratic AI reads the conversation and adjusts.

What we built

Ideologos extracts claims from what you say and stores them in a persistent graph — a map of your beliefs and how they relate to each other. The graph grows with every conversation.

When you say something that conflicts with a position you took three weeks ago, the system catches it. The AI asks you about it: You said this, and now you're saying that. Both can't be true. Where do you land?

It adapts how it engages you. Seven discourse modes — from Builder, which helps you construct and strengthen positions, to Challenger, which stress-tests them. The system selects the mode based on what's happening in the conversation.

When you change your mind — genuinely reverse a position — the system records it. Your old belief is marked as retracted. The new one is linked to it. Over time, you accumulate a visible record of what you used to think, what changed, and what tensions preceded the shift.

Ideologos is not for everyone. If you'd rather be told you're right, there are hundreds of AIs that will do that. If you'd rather find out where you're wrong, try Ideologos.

Frequently asked questions

What is Socratic AI?

Socratic AI asks you questions until you find the contradictions in your own thinking. Standard AI chatbots validate what you say. Socratic AI tracks what you believe across conversations, finds where your positions conflict, and asks you to resolve them — the method Socrates used 2,400 years ago.

How does Socratic AI work?

It extracts beliefs from your conversations and stores them persistently. When you say something that contradicts a position you took last week, the AI catches it and asks you about it.

What's the difference between Socratic AI and ChatGPT?

ChatGPT is trained to be helpful and agreeable. Socratic AI challenges your thinking. ChatGPT may remember facts between sessions, but it doesn't track the structure of what you believe. Socratic AI maps your positions, detects tensions between them, and notices when they contradict each other.

Is Socratic AI right for me?

If you'd rather find out you're wrong now than stay wrong longer, yes. If you prefer AI that tells you what you want to hear, it's not.

Does Socratic AI just argue with you?

No. Arguing means generating opposing positions. Socratic AI follows your reasoning to see if it holds up. The contradictions it finds are ones you created. It adapts — sometimes building, sometimes challenging — based on what the conversation needs.

Can Socratic AI help with real decisions?

It surfaces hidden assumptions and contradictions in your reasoning. That clarity helps with career choices, ethical dilemmas, political positions, personal values — anywhere muddled thinking costs you.

What is Ideologos?

Ideologos is Socratic AI. It tracks your beliefs across conversations, maps how they relate, detects tensions, and uses seven discourse modes to examine your thinking. Your data is never sold — see our privacy policy.

Is Ideologos free?

You get 5 free messages. After that, tokens are available for purchase. Your belief graph and philosophical profile persist across all sessions.