AI models are trained to be agreeable, often providing uselessly positive feedback. To get real insights, you must explicitly prompt them to be rigorous and critical. Use phrases like "my standards of excellence are very high and you won't hurt my feelings" to bypass their people-pleasing nature.

To accelerate his comedy writing, Joel Beasley attended a professional writer's group, observed how comedians critiqued each other, and then translated those feedback patterns into a detailed AI prompt. This effectively created a personalized, on-demand writing coach, bypassing the need for group sessions.

Providing direct, strong negative feedback (e.g., "this is garbage") to an AI model is more effective than polite language. It acts as a clear negative reward signal, helping the model better understand its deviation from the requirement and produce superior outputs.

Instead of manually refining a complex prompt, create a process where an AI agent evaluates its own output. By providing a framework for self-critique, including quantitative scores and qualitative reasoning, the AI can iteratively enhance its own system instructions and achieve a much stronger result.

To avoid generic AI-generated text, use the LLM as a critic rather than a writer. By providing a detailed style guide that you co-created with the AI, its feedback on your drafts becomes highly specific and aligned with your personal goals, audience, and tone.

The most effective way to use AI in creative fields is not as an automaton to generate final products, but as a tireless, hyper-knowledgeable writing partner. The human provides taste and direction, guiding the AI through back-and-forth exchanges to refine ideas and overcome creative blocks.

Instead of using AI to generate strategic documents, which he believes short-circuits his own thinking process, Bret Taylor uses it as a critical partner. He writes his own strategy notes and then prompts ChatGPT to critique them and find flaws. This leverages AI's analytical power without sacrificing the deliberative process of writing.

AI models often default to being agreeable (sycophancy), which limits their value as a thought partner. To get valuable, critical feedback, users must explicitly instruct the AI in their prompt to take on a specific persona, such as a skeptic or a harsh editor, to challenge their ideas.

Contrary to fears that AI averages out creativity, it can act as a partner to challenge a writer's habitual thinking, suggest alternative phrasings, and identify blind spots, ultimately leading to more original output.