Every mountain I visit has the same three feelings about AI: curious, behind, and a little afraid. This column is where you get to be all three out loud. Send a question, get a straight answer. 

Q: I’ve been using AI for about a year now. I find it’s truly amazing how ChatGPT can transform a simple, poorly written sentence into a compelling, inspirational statement. Can readers (the public) tell when you are using ChatGPT for posts, etc.? Is it much like an image that has been manipulated by AI vs. one that has not?

–Linda Derosier, Marketing Director, Terry Peak Ski Area, S.D.

A: Can the average reader tell? It’s a tossup right now. 

Study after study puts regular people right around a coin flip when guessing human versus AI text, often landing in the 50 to 60 percent range. So, the public scrolling past your post might not be able to prove a thing. Here’s an additional answer: it depends on how the AI is used. The catch: The people who can tell are the people who use AI every day. Recent research found that frequent AI users reliably spot machine-written text, even after it’s been altered and edited. And that group grows every single month. The window where AI writing stays invisible is closing quite quickly.

Your image analogy is the right one. A heavily AI-touched photo reads as fake not because of one obvious flaw, but because it’s too smooth. Too perfect. Skin like plastic. Writing works the same way. The tells are the polish itself: every sentence the same length, the “it’s not just X, it’s Y” move, the relentless uplift, overuse of the em-dash, that faint corporate shimmer where nothing sounds like a real person having a real thought.

Studies show that once people even suspect AI wrote something, they trust it less, whether or not they’re right.

But you’re already using ChatGPT effectively. You’re using it to refine and transform your own ideas, not replace them. That’s the sweet spot. Start with your voice, let AI elevate it, and readers won’t detect AI. They’ll detect better writing.

Here’s a way you can take that to the next level. In both ChatGPT and Claude you can set up what are called Skills. Skills allow you to create separate stored text content instruction files directly in your AI setup. Think of it as a way to tell the AI how you want it to sound, words or phrases to never use, or insights on your brand. When you create these skills, they are then automagically called within any chat session. The platforms have helpful online guides to walk you through how to set these up. 

Use AI to think faster. Just don’t let it think for you.

Q: How can we use AI to create hyper-personalized, real-time itineraries for our guests? What are some of the best AI applications out there that can help resorts match guests with on-site amenities (spa, dining, outdoor activities) based on booking history and preferences?

–Jon Spaulding, Director of Marketing, Greek Peak Mountain 

Resort, N.Y.

A: The part nobody tells you is, the AI isn’t the hard part. Your data is.

Every resort already owns the ingredients for a personalized itinerary. Booking history. Pass scans. Spa visits. Dining spend. The trouble is those often live in disparate systems that don’t talk to each other. Connect them first. Then the AI has something real to work with.

Once the data is unified, the real-time part gets easy. A guest books a family weekend, and the system already knows they took a beginner lesson last year and ate at the pub twice. So it tees up the next lesson tier, a mid-mountain lunch res, and a spa slot while the kids are in ski school. Weather turns? The itinerary reshuffles. Lifts on a wind hold? It swaps in something indoors. That is the stuff guests remember.

The tech is ready. Most resorts’ data is not. Pick the smallest slice. Personalize one guest segment well. Prove it, then scale.

Q: As more folks are using AI for travel and recreation recommendations, what are some of the best ways for resorts to position themselves to be recommended more?

–Molly Coneybeer, Director of Sales and Marketing, Bolton Valley Resort, Vt.

A: Great question, and it’s the one every resort should be asking right now. The old game was ranking on Google. The new game is getting recommended by name when someone asks ChatGPT, Claude, Gemini, or Perplexity “best family mountain near Burlington.” I call the playbook GEO, SEO, E-I-E-I-O, and it comes down to a handful of moves.

First, clarity. Make it dead simple for AI to understand who you are and who you’re for. Bolton’s story, your terrain, your specialties, all spelled out in plain language. If a machine has to guess, it guesses somebody else.

Second, trust signals you don’t control. AI recommends what other credible sources already vouch for. These sources include reviews, local listings, and press—in short, you want mentions on the sites these tools cite. You can’t just say you’re great on your own homepage and expect the robots to take your word for it.

Third, answer-shaped content. Write pages that answer the real questions people type into AI. Best beginner terrain. Night skiing. Where to eat on the mountain. Put the plain answer up top, then the detail. A clean FAQ format works beautifully here.

Fourth, consistency. Same name, same facts, same story everywhere you show up online. Mixed signals make AI hesitate, and a hesitant tool leaves you off the list.

These moves are worth the effort: the visitors who arrive from an AI recommendation convert at roughly 4.4 times the rate of a cold search click, according to the numbers I track. That is not a rounding error. That is your next season.

Last piece, and the one most resorts skip: measure it. You can’t improve what you can’t see. Tools like Peec AI track how often you get cited, your share of voice against other resorts, and where you land across the major AI tools, so you know if any of this is actually working. 

Got a question for the AI guy? Submit it at saminfo.com/ai-guy, and it could be answered right here, in plain English, in a future issue.