Chat Said: Can You Trust AI? What Happens When It Gets It Wrong?

Can You Trust AI? What Happens When It Gets It Wrong?

 

“Chat said…” I caught myself saying this yesterday.  But how can you trust AI?  It has gone from a tool we consciously use to something that feels like a trusted friend, researcher, travel agent, and occasional know-it-all.

When did Chat become a regular part of my conversations? Not just something I use to clean up an email or brainstorm a blog post, but a source I casually reference as if I’d just gotten advice from a friend who happens to know a little bit about everything.

Planning a trip?  Ask Chat.

Trying to decipher Italian bureaucracy?  Ask Chat.

Looking for a hike, researching a new piece of outdoor gear, or figuring out whether that little restaurant in a mountain village is worth the detour?  You guessed it.  Ask Chat.

ChatGPT is convenient, it’s fast, and it can be remarkably helpful.  But somewhere along the way, I started wondering: When did we become so comfortable trusting a machine to help us make decisions?  And what happens when it gets things wrong?

 

How We Have Come to Trust AI Tools

 

What used to involve multiple Google searches, scrolling through forums, reading reviews, and texting friends can now start with one question. In one box. On ChatGPT or in places like Google Gemini or Claude.  Why have AI tools become so mainstream?

Discovering ChatGPT and Learning to Love It

 

Several years ago, I discovered ChatGPT and learned how to use it.  I taught our team how to use it to create content.  I even wrote blogs to help others become familiar with using ChatGPT. To create social posts for Instagram and Facebook.  To create blog posts.  To build videos and graphic designs.  Can you trust AI to do almost anything these days?  We were certainly testing the limits.

By the way, if you haven’t read either of these blogs yet, bookmark them for later.  They will give you the basics and get you comfortable using AI tools!

• How to Use ChatGPT to Create Social Posts
• Using AI for Blog Posts: 5 Steps to Get Started with Content Creation

 

But Can You Trust AI to Replace Your Content Creators?

 

While everything seemed to be going well, I could see the writing on the wall.  If I could trust AI and use tools like ChatGPT and Claude every day, how long before my clients would use them to replace me?

Can you trust AI to replace your content writer?  Your graphic designer?  Your video editor?  The answer was, sadly, yes.

As expected, we then began losing clients who employed Top Rope Media solely for content creation.  In 2025, I had to let team members go because our client roster began dwindling.  This was not good.  My way of life, the foundation I built Top Rope Media on, began crumbling.  Creating content and driving a client’s storytelling was now something ChatGPT could do. And in seconds.

 

How AI Kills Your Website Traffic

 

During this same time, we began to notice our clients’ website traffic began to dwindle.  People were using Claude, Google Gemini, and ChatGPT to organize trips, get health advice, or build the best marathon training plan.  And they never left the AI tool.  Can you trust AI?  Even with your mental health and well-being?  The answer seemed to be a resounding yes.

Someone might ask an AI tool to recommend a cycling training plan, or build a training plan informed by a coach’s website, and never click through to that website.  The business’s expertise may have influenced the decision even though the visit never appears in Google Analytics.

In addition, we noticed new traffic coming to their websites from places like Pakistan. Singapore. New Delhi.  At first, I saw this as strange. Many of our clients only sell to the US market.  Therefore, none of our ads were targeting places like India or Indonesia.  Then it started to make sense.  The foreign hits to a client’s website could be coming from AI bots and website crawlers.  Alternatively, a visit from an overseas IP address could be a real person, but likely a VPN, a proxy, or automated AI activity.

As we began digging deeper, we discovered new reasons to loathe how AI was taking over.  I even wrote a full blog on How AI Search Tools Are Killing Website Traffic for Small Businesses and what to do about it.

 

Getting Referred to By ChatGPT

 

Then, a funny thing happened. I began getting referred by ChatGPT, Claude, and Google Gemini.  I would hop on an introductory call with a potential client and ask how they found us.  The answers were always surprisingly similar.

“I asked ChatGPT who the best marketing agencies in the outdoor industry were, and you were at the top of the list.”

“I asked Google to pull the best holiday marketing tips, and Google Gemini gave me tips from one of your blogs.”

“I asked Claude who could help us book our cycling tours and sell more trips, and Top Rope Media came up.”

The more I asked, the more I heard “Chat said…” “Google recommended.” “Claude referred me.”  It was as if they were real people, friends even, providing advice.

 

How to Make ChatGPT Work For You

 

For us, this was incredible.  Most of our clients had come to us through referrals from current and past clients.  We were now getting strangers to sign up for our marketing services.  All the content we had been developing for Top Rope Media- the blogs, the videos, the SEO work on the back end, the social posts- it was all paying off.

How do you do this for your business?  Well, the way people discover everything, from marketing tips to tour operators or outdoor gear recommendations, is changing.  Even if your business hasn’t caught up yet.  A traveler might no longer start with “guided hiking tours in Greece” on Google.  They might instead ask Claude, “Where can I find a small-group, multi-day hiking trip in mainland Greece that avoids crowded tourist destinations and has knowledgeable local guides?”

This type of search changes what your business needs to communicate online.  You need to explain your offering in enough detail that both people and AI-powered search tools can understand it.  Generic marketing copy is unlikely to answer the questions or get you referred.

I even had a new client pay me to teach them how to get referred to by ChatGPT!  I organized a 2-hour training session on how to build out their new website and create better content to be more easily discovered by AI tools.  This became a new offering for us, and if you are interested in setting up your own training session, drop us a note!

 

 

Can You Trust AI to Deliver Accurate Information?

The Benefits of Using AI

 

We began to work more in harmony with AI tools than ever before. With our client work, we were saving time, money, and resources we were being used more efficiently.  Chat could help us rewrite an email to sound more professional. Generate marketing ideas for our fall campaigns.

Outside of work, we were using ChatGPT to recommend hikes, plan our trips, and navigate the Italian citizenship process.  It was like speaking with an extremely knowledgeable friend who gave you the best recommendations. From where to have aperitivo to ride routes and things to do.  I stopped using Hotels.com and TripAdvisor.

And unlike searching on Google, Chat keeps all your searches neatly stored.  You can continue working in the same Chat, or start a new Chat on a different subject.

 

The Downside of Trusting AI

 

The downsides of depending on AI Tools like ChatGPT began appearing slowly.  And when Chat got it wrong, it was very disappointing.

We’d drive an hour to a mountain refugio that Chat highly recommended and said was open, only to find a closed-for-the-season sign hanging on its door.  Or when Chat said we should do this beautiful loop on the bike, only to realize the road wasn’t fully paved, and it involved more elevation gain than we anticipated.

In 2025, I began seeing Instagram Reels of people hiking up via ferratas near where we live in Italy, the famous ‘iron ladders’ of the Dolomiti, with their mountain bikes.  It was hard to tell if these were real ascents or generated by AI.  Then word spread that ChatGPT was actually recommending and providing instructions on how to do this. This prompted the Rotaliana Alpine Rescue team to issue warnings. “If ChatGPT recommends cycling on the via ferrata…”

Taking your bike on a via ferrata was absurd!  I’ve hiked some of these ferratas, and the techniques required to summit do not include shouldering a 30-pound mountain bike.

 

Can You Trust AI for Travel and Trip Planning? You Can, But It Can Be Extremely Dangerous.

 

Then, the news got worse. In September 2026, I read an article about three novice hikers who were rescued trying to summit Mt. Shasta in California, after using Google Gemini to plan their ascent of the 14,000ft. volcano.  In October, a 16-year-old was rescued by a helicopter from a technical cliff known as the “Widowmaker” in British Columbia after using Claude to map his route.

The final nail in the coffin was when we were researching our recent trip to Sardinia.  Tyler used Claude to build out our trip, which was focused primarily on road cycling and rock climbing.  Claude recommended that our first stop should include rock climbing at Cala Spinosa on Capo Testa.

I pulled up images of Cala Spinosa and Capo Testa on Instagram and Facebook; it looked absolutely gorgeous! Climbers ascending stunning granite routes right above the sea.  I then looked at our climbing app, Vertical Life, and pulled up sites like Climbing Sardinia, the Crag, and Planet Mountain to learn more about the climbing there.  Turns out local climber and guide Maurizio Oviglia and the Climbing Sardinia Association have issued urgent, high-risk warnings pleading with people NOT to climb at Capo Testa. That despite what you might see on Facebook, YouTube, or Instagram, it is extremely dangerous.  The old stainless steel bolts were rotting from the inside out due to exposure to the heat, humidity, and the sea air.  Bolts could fail without warning.

Um, okay. No thanks, Claude.  I reorganized our trip.

 

Final Thoughts: How Much Can You Trust AI?

 

AI should be your starting point, not the final authority.  Do not let it make every decision for you.  Especially when it involves risky adventures.  Even an easy walk in the park can be dangerous if your AI-generated plan leads you down the wrong trail or suggests a route will take you two hours when it actually takes six.

Sometimes the unexpected discovery, the wrong turn, or the conversation with a local becomes the best part of the experience.  Sometimes it doesn’t.

By trusting AI, we’re trading some of the process of discovery for convenience.  By all means, use your friend ChatGPT to explore.  Brainstorm, compare, summarize, and build a first draft.  However, don’t forget to check the sources. Confirm schedules, closures, pricing, and availability.  For technical outdoor safety, medical questions, and other high-stakes decisions, use qualified people and authoritative information.

The same can be said for trusting AI for your business needs.  Question things.  Talk to people who know their stuff.  Always be learning.  Reading.  Figure something out the hard way.

Can you trust AI?  Here are the blogs I’ve referred to help you discover it for yourself.

• How to Use ChatGPT to Create Social Posts
• Using AI for Blog Posts: 5 Steps to Get Started with Content Creation
• How AI Search Tools Are Killing Website Traffic for Small Businesses

 

Thanks so much for reading.  Or asking Chat to read this for you,

Meredith Caccese McConvill, Top Rope Media