What if the trap works in both directions? The lawyer who opens with a specific framing anchors the AI on that frame. The AI optimizes within it. The lawyer reads the output as validation. Neither side surfaces what was outside the frame to begin with.
The recommended prompt is a great practical take on the first draft trap. But I keep thinking about the upstream anchor: the lawyer's own description of the matter is itself a frame, and the three strategic alternatives the AI generates will be shaped by how the question was framed.
I like your prompt. Another idea is to use old fashioned free writing to help get out your thoughts and then let AI synthesize them for you. I'll pop in a half-baked idea and then converse with the AI to flesh it out.
What if the trap works in both directions? The lawyer who opens with a specific framing anchors the AI on that frame. The AI optimizes within it. The lawyer reads the output as validation. Neither side surfaces what was outside the frame to begin with.
The recommended prompt is a great practical take on the first draft trap. But I keep thinking about the upstream anchor: the lawyer's own description of the matter is itself a frame, and the three strategic alternatives the AI generates will be shaped by how the question was framed.
I like your prompt. Another idea is to use old fashioned free writing to help get out your thoughts and then let AI synthesize them for you. I'll pop in a half-baked idea and then converse with the AI to flesh it out.
Another thought-provoking proposition causing me to rethink and hopefully improve my awareness when interacting with AI going forward. Thanks