Are you tired of bro marketing, manipulative algorithms, and sales funnels that encourage you to ‘trip’ a prospect to make her buy more? Then you’re in the right place.
On Small Stage, Big Impact veteran digital strategist, Renia Carsillo, combines solo teaching episodes with interviews to help you learn how to build a sustainable, people-first digital strategy that works for your business. Along the way, she speaks to digital activists, upstart developers, badass content creators, brave community leaders, and other behind-the-scenes folx you probably haven’t heard on a podcast before. To help you answer one important question:
How do we do better digital (#DoBetterDigital) so we can be better humans?
Google’s AI Overviews are confident, familiar, and wrong more often than you’d think. In this episode, I talk about why I stopped trusting them blindly, and why most of us need to question our search habits, myself included.
Most of us are bad at searching and don’t know it, and because we expect Google to be accurate, we trust it more than other LLMs. That’s why I’ve started trusting Reddit, Discord, and actual humans over Google.
In this episode, we’ll get into:
- Why using Google every day doesn’t make us good at searching.
- How AI Overviews can sound convincing and still get the answer wrong.
- What inaccurate AI summaries mean for small businesses trying to be found and trusted.
- Why I’m turning to Reddit, Discord, and actual humans for advice grounded in experience.
Google’s AI Overviews are confident, familiar, and wrong more often than you’d think.
At least one evening a week, I have the same small-stakes fight with my husband. He Googles something and points out that the sources are cited. I remind him that doesn’t mean they’re real.
Most of you (and even me at least half the time) would agree with him. For the majority of searches, Gemini works well enough. Except when it really doesn’t.
In today’s episode, I talk about why I stopped trusting Google’s AI Overviews blindly, and why I’ve started trusting Reddit, Discord, and actual humans instead.
In my mind, Google in 2026 is about as trustworthy as it was in 2010.
The first time we had this argument, we were curious about the number of University of Kentucky graduates. His quick search said around 300,000 to 400,000. That’s not possible when there are only around 650,000 people in Lexington. Even though I was logged in and searching from Lexington, Gemini gave us an answer about the United Kingdom because of the way we worded the query.
Google used to be a starting point, not the endgame. Over the last 20 years, we’ve come to rely on it as the source of truth.
Google is part of our daily lives in a way that other LLMs are not.
We’re skeptical of ChatGPT and Claude because they’re newer. Google is familiar, so we trust it automatically. It’s called the familiarity heuristic, and Alphabet (Google’s parent company) is relying on it to keep us using their product at the expense of what is reliable or true.
Most of us have no idea how to search, even when we use search features on a regular basis.
When new team members start working with Realign Consulting, we require them to train in proper Google query techniques. None of this has ever been intuitive, and that problem has compounded a thousandfold since AI Overviews became the default result for most search queries.
Simply and clearly put, most users are terrible at using Google search properly, even when they use it every day.
In the above example, even the proof the search results offered was ungrounded. The source the AI Overview cited didn’t say what the overview claimed it did. And the oft-cited statistic that AI Overviews are around 90% accurate doesn’t account for users who query poorly.
These days, I’m Googling a lot less for answers. I’m spending more time on Reddit, on Substack, in my Discord servers, and (*gasp*) actually texting my mentors, friends, and colleagues.
What we Cover
- Why Gemini overviews are only as good as your query, and why they’re sometimes wrong even when your query is fine
- How a simple search about University of Kentucky graduates turned into an answer about the entire United Kingdom
- Why we trust Google more than ChatGPT or Claude, and what the familiarity heuristic has to do with it
- Why we train new team members at Realign Consulting in proper Google query techniques
- What an “ungrounded” AI Overview is, and why the 90% accuracy statistic leaves out people who query poorly
- How Gemini can step in even after someone lands on your website, and what that means for small businesses
- Where I’m going for answers now: Reddit, Substack, Discord, and the actual humans in my life
Episode Recap
- [00:00] Why I’m trusting Reddit, Discord, and actual humans over Google, plus a welcome to this season of Small Stage, Big Impact
- [00:54] Why AI Overviews are only as good as your query, the recurring fight with my husband, and the University of Kentucky graduates search
- [03:17] The corrected query reveals an answer about the United Kingdom, and why Google in 2026 feels about as trustworthy as it did in 2010
- [05:34] Library research, the guy in the JNCOs, and why a Gemini-driven AI Overview is different from ChatGPT or Claude
- [07:45] Why Google’s reach makes this a bigger issue, and how we started training team members in proper query techniques
- [10:10] The ungrounded AI Overview about how well people search, the 90% accuracy statistic, and why experts say to talk to the LLM instead of typing
- [12:31] Gemini on your own website, Business Insider’s reporting on small businesses, and the end user’s own unreliability
- [14:55] Texting mentors and friends for referrals, the bartender who knows the best plumber, and closing thoughts
Resources Mentioned in this Episode
- The original Substack article
- Google query guide — covers basic steps for properly formulating a query and includes a QR sheet for using operator keys
- Big Tech’s Continued War on Local Business — last week’s piece on how AI Overviews and overzealous ad models have made purely organic traction unlikely
- Business Insider — reporting on how inaccurate AI Overviews are sabotaging many small businesses’ digital marketing efforts
- The 6% Project
Sometimes our show notes contain affiliate links. We only recommend books, products, or services we feel great about and believe will support you in your work. These small commissions help pay for the production of our show each week. We do not accept show sponsorships or host guests in exchange for compensation of any kind. Thank you for supporting us by using our links to purchase when we’ve recommended something that piques your interest.
