AI as a Peer Review. I’ve been exploring AI use cases that…

I’ve been exploring AI use cases that don’t focus on efficiency, but on thinking. This is a three-pass AI method I’ve been using to break confirmation bias and think bigger.

So much AI chatter is around efficiency, or “cost-out” — but I think the bigger unlock is its ability to help you think critically, more expansively, or to look for things you’re either information-restrained from seeing, or where you have bias, or let’s face it, fatigue, or even subconsciously not seeing something.

The most straightforward “peer review” hypothetical I have is working on an investment thesis. I am not yet ready to fully automate this with an AI Agent, although I’ve heard some are attempting to do so. That said, once I am 60–70% done with my own work, I’m certainly not afraid to get “fresh eyes” on what I’m doing and thinking. Here are three examples of what I’ve been asking AI to do (this is assuming ChatGPT):

1. Critique my draft. Ask for agreements, disagreements, and open questions. “Deep Research” mode (a setting that prioritizes gathering and synthesizing long-form sources) works well for this.
2. Rerun it with no web. I like to get a read that has more control against bias and the groupthink that can come from the web. Toggle off “Web Search” in personalization settings, and “Thinking” mode works well here.
3. Build its own thesis. Don’t feed mine; see what emerges independently. “Deep Research” works well here.

Reading those three passes side-by-side helps me confirm or kill ideas, find blind spots, and confront confirmation bias.

Want to try it?

1. Finish your work to 70%.
2. Prompt: 1) Critique it, 2) reanalyze with the web off, and 3) generate an independent, comparable analysis.
3. Review. Where did it disagree? What risks did you downplay?

Let’s face it — we can all benefit from a peer review and a safe space to have our work critiqued. I’m not ready to hand AI the keys, but I certainly will give it a flashlight!