Edwin Mode
Ask AI to Improve Your Thinking
Aside from my immediate family, my favorite relative was Uncle Edwin. He served in the Navy for 25 years and then became a high school math teacher in Beaumont, Texas. Among other things, he coached the school’s slide rule team. He was witty, well-read, and a fine thinker.
He also disagreed with almost everything I said.
If I said something was black, Edwin would argue it was white. If I said it was white, he’d argue it was black. He didn’t argue angrily. He just loved a good debate. For him, an argument wasn’t about winning and losing. It was about thinking together.
And it worked. Edwin made me defend my ideas. He poked holes in my reasoning. He forced me to sharpen my arguments. He taught me to think more clearly.
Edwin is long gone. But now we have ChatGPT. The difference is that Edwin taught me how to think. ChatGPT does my thinking for me. ChatGPT is often very helpful. But one thing it doesn’t do is teach me to be a better thinker.
ChatGPT has one weakness. Unlike Edwin, ChatGPT always agrees with me. It’s too polite. So I’ve started using a simple trick I call Edwin Mode.
Whenever I want ChatGPT to challenge my thinking instead of thinking for me, I begin my prompt with the words:
“Edwin Mode.”
This tells the system to stop being helpful and start being skeptical. It should question my assumptions, probe my logic, suggest alternatives, and point out weaknesses. In short, it should behave like Uncle Edwin.
But different ideas, at different stages of development, need different challenges. Sometimes you just want a gentle nudge. Other times you want your logic stress-tested.
So I use two versions of Edwin Mode.
Gentle Edwin Mode
I use this when my idea is still forming and I want thoughtful questions rather than a full-on debate.
Prompt
Edwin Mode (Gentle): I’m exploring an idea. Ask thoughtful questions that might help me clarify or strengthen it. Point out assumptions I may be making and suggest angles I haven’t considered. {Fill in the idea here.}
This is like Edwin leaning back in his chair and saying, “Are you sure about that?”
Full Debate Edwin Mode
I use this when I think my argument is fully developed and I want to test it.
Prompt
Edwin Mode (Debate): I’m going to present an argument. Challenge it as a skeptical but intelligent critic. Identify weak assumptions, logical gaps, counterexamples, and alternative explanations. Treat this as a friendly debate intended to strengthen the idea. {Fill in the idea here.}
This is the version where Edwin rolls up his sleeves and makes you sweat.
Please try these out in your work and let me know the results.
The main idea here is not mine. I picked it up from Tom’s Guide (click here), where it’s called “potato” mode. I’ve attached it to my Uncle Edwin. I hope I’ve also added value by incorporating both “Gentle” and “Full Debate” modes. I also tip my hat to Tom’s Guide for sharpening my thinking.


This just melted the scales from my lizard query eyes! I tell AI to “make it better” or “give me more”, but have not trusted it to parse more nuanced constructs.
Thanks Travis. Your thoughts reminded me of a line from an article about AI several years ago written by 3 authors, including Henry Kissinger. The line was essentially “AI will provide more answers but less understanding.”