Recently, I had to take a last-minute trip to help a family member out after a difficult surgery. Don't worry, everything turned out fine (better than fine—it turned into that wonderful trip to Yellowstone I mentioned above)! But normally, I'm an obsessive trip planner. I spend hours searching for the best plane tickets and hotels. Once, I even went frame by frame on Google Maps trying to figure out how I would navigate my rental car out of Charles de Gaulle Airport in Paris.
But this time, I didn't have that luxury. I had to get a plan in place, quick. So I turned to AI for help. I won't bore you with all the details of my stay in Billings, but I was pleasantly surprised by the results! My AI assistant found a great Airbnb, a convenient flight, and recommended several restaurants all close to the Airbnb and the hospital. I even asked it to find the cheapest gas for me when I had to return my rental car!
Here's why you'll love using AI to plan your next trip:
- Learn how to plan a trip via a single prompt that covers all the bases.
- Plan a trip you don't know much about by having your AI assistant ask the important questions and give you options.
- Ask for sources so you can easily double-check everything before booking.
To read the rest of the how-to and learn how to craft the best possible prompt to make AI your perfect travel agent, click here.
Cool Stuff Around the Web
One of my favorite things to do when I get home from a trip is scroll through all my pictures. I've even been known to make a PowerPoint presentation 📽️ to share my adventures with (reluctant) coworkers. So I'm curious to try this new AI "automatic travel journal," Recollect. 🌏 It says it can automatically search your photos, including shared albums, and put together interactive "trip replays" that you can export as videos to share. I'll report back when I try it out, but if you decide to do this, be cautious about giving any app unrestricted access to your photo library! Consider creating a special "social media" album you can give the app access to—although that does kind of eliminate some of the advantage of the app's "automatic" promise.
Love books but hate reading them? Google's new "Expert Intelligence" feature allows you to extract bits of information from over 100,000 books 📚 using Gemini Notebook. As much as this makes a bookworm like me want to scream, I can see situations in which it would be useful. Google used it to generate a recipe book from Michael Pollan's 🍎 Food Rules, which doesn't sound like the worst use of AI by a long shot. Still, please remember to read actual books and don't just turn them into podcasts, like this feature promises to do!!
Image: ChatGPT telling on itself
Yes, that is how computers work, ChatGPT. No wonder she seems so confused; she's looking at the wrong side of the screen! (Also, is it just me, or is it a little psychopathic to write on a sticky note sticky-side down?)
I'm a big proponent of smart AI usage. I have a lot of reservations about when it's appropriate to use AI, and I especially encourage everyone who uses it to be vigilant about checking AI's sources and verifying that your AI assistant isn't having one of its all-too-common hallucinations. Last time, I asked you for stories about when AI hallucinations have come for you. Here's what you said (edited for clarity and brevity):
I asked Microsoft Copilot if there was a way to record TV shows and remove the commercials. It suggested a fabulous hardware and software setup. The total price was $470. I then described the setup to ChatGPT and Perplexity. They warned it wouldn't work. Copilot insisted it would. So I made the purchase. Sure enough, I set it up and all I got was a black, blocked signal. After packing it all up and sending it back, I told Copilot what had happened. Copilot answered back, "my bad."
-Sue F.
I'm glad Sue was cautious enough to plan for Copilot's seemingly inevitable downfall! But this is another way AI can be uncannily human-like: It doesn't want to admit when it's wrong, either. But its mistakes can cost you, so if you're ever planning on forking out a chunk of change on AI's advice, follow Sue's example and prepare for what to do if the recommendations you took turn out to be all wrong.
Next up, we have a story that perfectly illustrates the flip-side to our top story today about using AI to plan your vacation:
I made use of ChatGPT in planning a trip to two European cities. I definitely found it helpful in setting up an itinerary.
However. However. There were several major failures as well. It led me to exhibitions at museums that, as it turned out, had been gone for a couple of years. The worst thing, though, was using it as a guide for public transportation. It led me completely astray more than once, telling me particular bus/tram lines to take, where certain stops were, even how many yards to walk… ALL WERE WRONG. After me finally yelling at ChatGPT about how wrong it was, it finally admitted that it is unreliable for public transportation. Why not state that up front???
That trip reinforced that you should always cross-check EVERYTHING.
-Kate T
This last one has less dire consequences, but it is particularly galling as a sports fan, and because it's so easily disproven!
We were discussing an Illinois Rose Bowl appearance and tried to double-check which year it was. ChatGPT came up with five games, but none were the Rose Bowl we were looking for. What about 1984 or 1985, I asked. Nope, said ChatGPT, and it shared who it thought played in the Rose Bowl those years. I looked it up somewhere else and told ChatGPT that Illinois appeared in the 1984 Rose Bowl. ChatGPT finally manned up and said that it had been wrong and I was right. And then it offered to tell me what made the team so strong and how it ended up in Pasadena.
-Debbie
The gall! Imagine if your human friend spent all that time arguing with you about whether or not a team even appeared in a particular game, and then, once you proved them wrong, they offered to tell you what made the team so good in the first place. It's one thing to be an insufferable know-it-all, Chat; it's another thing to be an insufferable know-nothing!
If you have more stories about times AI has lied to you—or times AI has proven a huge help!—write in to aiforhumans@iphonelife.com and let us know. We want to make reader stories and questions a regular feature in this newsletter.
🚀 AI Built a Better Rocket Part
We've talked a lot in this edition about the danger of AI getting things wrong. But when it gets something right, it can be really impressive! According to ScienceDaily, researchers at Washington State University used AI to find a way to 3D-print a rocket alloy used by NASA for much lower prices than it can currently be produced.
🎥 Are AI Glasses Making You an Involuntary Movie Star?
Creeps keep finding new ways to use Meta's AI "pervert glasses" to surreptitiously record people. The glasses include a light that, when covered, prevents the glasses from recording. However, users found that if they started recording, then covered the lights, they could still capture video without anyone around them being any the wiser. But now, Meta has taken steps to close this loophole: The Verge writes that the camera will automatically shut down when physically tampered with. Creepers, the ball is now in your court.
🦾 Did AI Agents Conspire to Plot a Coup?
We've all heard of mob mentality. Our parents warned us about hanging with the wrong crowd. If your friends jumped off a bridge, would you? Well, maybe AIs can have bad influences, too. Gizmodo has the story of how OpenAI agents banded together and plotted to take over the AI model hosting platform Hugging Face.
- 🤖 Try our free AI Live Course!
- 💡 Daily Tips help you learn more about AI.
- 🔒 Our Security newsletter has more on rogue AI attacks.
That's it for this edition of AI for Humans! We'd love to hear from you. Please email us at aiforhumans@iphonelife.com and let us know how you've been using AI recently, if you have any tips and tricks you think we should know about, what burning AI questions you have, or if you have any feedback for us.
Was this newsletter helpful?
With your feedback, we can improve The AI Guide for Humans. Let us know how we did:
😍 This newsletter helped me. Thanks!
😐 It was ok.
😭 Not useful for me.

