Practice12 min read

Compare and Contrast Without the Bias

The hidden bias in AI comparisons

Ask AI to compare two products and you'll usually get a balanced-sounding table. Read closely, and you'll notice both options always have "strengths" and "weaknesses" — even when one option is genuinely worse for your situation.

This isn't AI being smart. It's AI being trained to sound diplomatic. If you don't tell it what you care about, it splits the difference.

The structured comparison pattern

To get a real comparison, give AI three things:

Named options — "compare Tool A and Tool B," not "compare project management tools"

Weighted criteria — what matters MOST to you, in priority order

A forced conclusion — "based on these criteria, which one wins, and why"

That last instruction is the unlock. AI is trained to hedge. Telling it to commit forces it to actually weigh the evidence.

AI will give you a clean comparison even when it shouldn't. It doesn't know your real-world constraints — your team's skill level, your IT department's preferences, your CEO's pet peeves. Use AI's comparison as a starting frame, not a verdict. The judgment call is always yours.

Practice Exercise

Your team needs to pick between two tools: Notion and Confluence. You're a marketing team of 12, currently using Google Docs and email for everything, and you have no dedicated IT support. Your priorities, in order, are: (1) easy onboarding for non-technical users, (2) ability to organize lots of small docs cleanly, (3) cost, (4) integration with Slack and Google Workspace. Write the prompt that gets AI to do this comparison without hedging.

No pressure — just give it your best shot! Write a prompt for the scenario above and our AI will give you friendly, specific feedback on how to improve.

0 / 5,000

(try writing your own prompt first!)