He Asked ChatGPT to Replace 200 Customer Interviews. Here's What It Can't Tell You.

The Promise: AI Customer Interviews at Scale

Simon suggests creating ChatGPT personas — a 35-year-old marketing manager, a small business owner in Ohio — and interviewing them about your product idea. He claims this replaces traditional market research that costs thousands.

Why This Falls Apart

ChatGPT personas are consensus machines. They reflect the average of their training data, not real market signals. Ask a ChatGPT persona if they'd pay for your SaaS product and they'll almost always say yes — because the training data skews toward optimism in business contexts. Real customers say no. A lot. That friction is where the actual insights live.

What Actually Works

Perplexity AI gives you real-time market data with cited sources. Instead of asking a fake persona what they think, you can search for actual customer complaints, competitor reviews, and market gaps — with links to verify every claim.

ChatGPT Deep Research goes deeper: it runs multi-step research across dozens of sources, synthesizes findings, and shows you the actual competitive landscape. Not what a persona imagines the landscape looks like.

Consensus searches across 200M+ academic papers. If you're validating a health, education, or B2B concept, it finds peer-reviewed evidence for or against your thesis. Try getting that from a ChatGPT persona.

The Real Play

AI personas are useful for brainstorming copy and testing messaging tone. They're terrible for validating whether a market exists. Use Perplexity for real-time competitive intelligence, Deep Research for comprehensive market analysis, and Consensus when you need evidence-backed validation. The 20 minutes you'd spend setting up ChatGPT personas is better spent on tools that show you what real people actually said.