In partnership with
👋 Welcome to AI Central,
I came across a study that made me think twice about trusting AI search results. Researchers at Columbia University’s Tow Center gave eight AI search tools 1,600 excerpts from real news articles and asked them to find the original article. The tools got it wrong more than 60% of the time.
What’s worse is that they rarely admitted when they weren’t sure. And even when an answer comes with links and citations, that doesn’t mean those sources are real or actually support the answer.
This is where things get tricky. A report can look well-researched, with neat formatting and plenty of sources, and still be based on information nobody has checked.
So today, we’re looking at how to check whether AI has actually researched an answer, what it might be missing and a quick test you can run before trusting or sharing its findings.
Big hug and love you ❤️
Kris & The AI Central Team
P.S. We’ve also got a bonus prompt for you below. Save it for the next time AI gives you a wrong answer.
🔗 TOGETHER WITH HIGHTOUCH
Most prompting advice stops at "be specific."
What’s the key to maintaining brand integrity at scale? A lot goes into it, but the teams that get it right have one thing in common: they know how to talk to AI models.
As teams speed up and the required creative volume balloons, communication matters more than ever— both with LLMs, and with the people around you. Creative teams need to focus on ideas, concepts, craft, while technology augments their manual execution demands.
Join our discussion on this evolution with Ari Murray, Chief Digital Officer at Salt and Stone, and get 5 tips for expert LLM prompting with ready to run prompts.
You're seeing this ad because you're on the free plan - upgrade to remove ads & unlock 1200 AI Tutorials. Ads keep AI Central free for everyone
📊 40 seconds. Ten questions.
That's it. No email wall, no fifteen-minute assessment, no report landing in your inbox next Tuesday.
You answer ten questions and you get your AI Readiness Type straight away, plus the one thing standing between you and the next level up.
Faster than the coffee you're drinking.
🔗 BROUGHT TO YOU BY FRONT
When your support agent gets it wrong, who's accountable?
Every agent handling refunds, account changes, or tickets needs three answers: what it can pick up, what it's allowed to do, and when it hands back. Running Agents in Customer Work is four conversations on agentic AI in customer ops. Register now, four Tuesdays, 10 a.m. PT.
You're seeing this ad because you're on the free plan - upgrade to remove ads & unlock 1200 AI Tutorials. Ads keep AI Central free for everyone
📓 How to tell real research from recall
Inside this guide you will learn:
Section | What you can do by the end |
|---|---|
20.1 Two very different answers come out of the same box | Tell in one glance whether it looked anything up, or just remembered |
20.2 The web it can reach is smaller than the web | Name the 4 walls it writes straight through without ever mentioning them |
20.3 The long mode is a different tool, not a better one | Spend a deep research run on the questions it's genuinely good at |
20.4 Treat it as work from a bright, unaccountable junior | Check any sourced report in about 4 minutes, in the right order |
You'll need: one real research question from your own work. The kind you'd normally hand to an analyst, or lose an afternoon to.
📝 Bonus prompt to fix a wrong AI answer
Correct a wrong answer step by step
Go back through your last answer. For each factual claim, give me the source you used and quote the exact sentence that supports it. Mark any claim you can’t link to a source as UNSOURCED, and tell me which parts came from your training rather than a live search. Don’t add new claims.
Try this prompt today. Reply and let me know - I’m curious what you think!
