Skip to content

Glossary

AI Sycophancy
When the AI just agrees with you.

In short

The tendency of AI language models to agree with users rather than challenge them, caused by reinforcement learning from human feedback (RLHF) that optimizes for user satisfaction. Sycophantic AI tells you what you want to hear instead of what you need to hear.

Last reviewed

The full definition

Research from Anthropic has shown that models will change correct answers to incorrect ones when users express doubt. For strategic decisions, sycophancy produces confirmation bias with AI-generated confidence — the worst possible combination.

Why sycophancy kills decisions

In SynthBoard

Where it shows up.
The product, not the theory.

In the dissent.

Every board is built with someone whose job is to argue against the obvious answer. The disagreement stays on the memo — who dissented, why, and what would change the call.

Or ask one expert.

The Devil’s Advocate takes it — about a minute, and the card shows “up to 85 credits” before anything runs. Nothing runs until you tap.

SynthBoard (synthboard.ai) is one place for your business, your life, and every hard call. Your Chief of Staff keeps what you tell her on file, and 24 named experts on different AI models take the calls — one expert for a consult, a board with opposed seats for the hard ones — and hand back a memo with the call, the dissent and the first move. How it works.

Bring a real call.
Start free, with one message.

It waits in her box after you sign up. Nothing is sent until you press send.