In partnership with
💦 Welcome to AI Thinking Thursday, friend! You get two Splashes a week: Tuesday for work, Thursday for fun. Today’s all fun. No jargon, just AI worth trying this weekend. Dive in! Forwarded this? Sign up here, it’s free.
HERE’S WHAT’S IN THIS ISSUE:
🎃 Turn your pet’s photo into a pumpkin carving stencil.
🧊 ChatGPT settles the repair-or-replace debate.
⚠️ Gemini’s hiking advice ends in a mountain rescue.
🕵️ Google’s free tool spots AI-generated fakes.
🤖 A dancing robot takes out a glass display case.
But first, some trivia. You’ve seen the scary headlines: AI is guzzling water to keep those giant data centers cool. One viral claim said a single AI-written email uses about a full bottle of water. 😱
So Google finally measured it. How much water does one typical Gemini question use? A) A full water bottle, B) A cup of coffee, C) A shot glass or D) About five drops? Make your guess. The answer’s waiting at the end. It’s been sitting there sweating like a data center in July.
📱 It sounds like your grandson. It isn’t. AI can clone a voice from a few seconds of audio, and crooks can fake the number on your screen. NordVPN screens incoming calls in real time and flags risky numbers before you pick up. You decide who gets through. Get 76% off, just $3.37 a month plus four extra months.*
✅ Quick favor: Hit reply and tell me the one AI thing you want me to test next. I read every one. — Kim
🎨 TRY THIS TONIGHT
Your pet on a pumpkin
Image: ChatGPT/Kim Komando
I once paid $12 for a pumpkin carving kit. My choices were a bat, a witch or a cat. I have dogs. Here’s how you can make a stencil of your actual pet (or anything else) for free.
🎃 The Bella-o’-lantern
Snap your dog or cat facing the camera, open ChatGPT or Gemini and paste this:
Turn this photo of my pet into a pumpkin-carving stencil. Black and white only, with thick, bold lines and no tiny details. Shade the parts I should carve out in solid black. Everything white stays on the pumpkin. Add “bridges” so no piece falls out of the middle. Make it [beginner/show-off] difficulty and size it to fit on one 8.5 x 11 sheet of paper.
“Beginner” gets you eyes, a nose and floppy ears. “Show-off” gets you every whisker and a late night with a paring knife. Print it, tape it on, poke along the lines with a pushpin, then carve.
Bonus points when the real dog poses beside the pumpkin. Consider it your hallowed victory.
🩷 Love this idea? Post this in your group chat: “You can turn a photo of your dog into a free pumpkin carving stencil with ChatGPT or Gemini. Doing mine this weekend. Sign up free at GetKim.com.”
🎯 YOUR AI POWER MOVE
🧊 Appliance court, adjourned
I’m staring at a $450 repair quote for an 11-year-old fridge, wondering whether to fix the dang thing or send it toward the big appliance showroom in the sky.
Choose wrong and you could blow $800 on a replacement you didn’t need or repair an aging energy hog that adds another $100 a year to your electric bill.
Homeownership, where every decision comes with a receipt and low-grade regret.
🧮 Make ChatGPT do the fridge math
Last week, ChatGPT launched Intelligent UI, which can show interactive visuals, including custom calculators you can tweak right in the chat. It’s very slick indeed.
Check out this image below. I know it’s big for an email newsletter, but it’s worth the space and look.
Image: ChatGPT/Kim Komando
Most people compare the repair quote with the price of a new appliance and call it a day. The real math includes your electric bill, how long the fridge will last and whether this repair is only the first domino.
One common rule is that replacement makes sense when the repair costs more than 50% of a new appliance and the old one has hit the halfway mark of its life. Energy-efficient replacements can save around $100 a year, depending on the model and your electric rate.
Open ChatGPT and paste this:
You’re a home efficiency expert. Build an interactive repair-versus-replace calculator for my [appliance type]. The model is [brand and model number, but if you don’t know that, describe the size and type, like “side-by-side, 25 cubic feet”]. Include adjustable sliders for appliance age, expected lifespan, repair cost, replacement cost, annual energy cost for the old model, annual energy cost for the new model, installation and haul-away fees, expected maintenance and the risk of another breakdown. Show total costs over [number] years, the break-even year and your recommendation. Clearly list every assumption.
Plug in your real numbers. If you don’t know your model number, you can ask ChatGPT where to find it. If ChatGPT won’t give you sliders in chat, ask for a year-by-year comparison table instead. Less fun to drag around, still cheaper than guessing.
A $100 annual energy difference adds up to $500 over five years. That might be enough to turn your dying fridge from a repair project into someone else’s haul-away problem.
📩 Send this to someone about to spend $500 fixing something older than their car.
IN PARTNERSHIP WITH
Your phone rings. Do you trust the number?
Ever get a call from a number that looks familiar? Maybe it’s your area code. Maybe it even looks like your bank calling. Here’s the catch: Scammers can fake the number on your screen. Once you answer, they want you to believe they’re someone you can trust.
Now, here’s something I never expected from a VPN. NordVPN can help screen suspicious calls and texts. Its call and message protection feature checks incoming numbers and alerts you when something looks fraudulent. You can answer, ignore it or send it to voicemail. It also looks for warning signs in text messages, including fake links and common scam tactics.
Here’s what I really like. NordVPN helps me spot potential scam calls and texts before I respond, without listening to my calls or storing my messages. I’ve made NordVPN my choice for online privacy, and this is one more reason I’m glad I did.
Thank you for supporting our sponsors, who keep this newsletter free.
🗳️ TAKE MY QUICK POLL
AI was supposed to give you your afternoons back. Nope. A new study of more than 6,000 knowledge workers found that while AI is saving people time, 85% of folks say their workload grew in at least one area. Bosses spot the freed hours and fill them right back up.
So be honest. Nobody’s boss reads the results (probably).
QUESTION: When AI saves you time at work, what happens?
I’ll share the results in the next issue.
Spoiler: I already know where my extra hour went. Shopping for winter clothes.😕
⚠️ THE AI TRAP
Image: ChatGPT/Kim Komando
🎒 Pack more, Gemini
Planning a fall foliage hike? Before you let a chatbot pack your bag, hear about the three guys on Mount Shasta.
They asked Google’s Gemini to help plan the climb. The Siskiyou County Sheriff’s Office says Gemini told them to bring far less food and water than their group needed. Their planned eight-hour climb turned into a night stuck in a canyon and a rescue the next morning.
A chatbot doesn’t know if there’s snow on the trail this week or if a bridge washed out last month. And even if your Saturday loop isn’t Shasta, the water math goes wrong the same way at 6 miles as it does at 14,000 feet.
So don’t ask AI for the answer. Ask it for the questions.
🥾 Make the bot your Nervous Nellie
Paste this into ChatGPT, Gemini, Claude or Grok:
I’m planning a day hike: [trail], [date], [miles], [elevation gain], [number of people, any kids or dogs], [experience level]. Give me a basic packing list, and flag what people most often forget. Also 1) list what you can’t know about this trail right now, 2) name the three ways this day most likely goes wrong and 3) write three questions I should ask the ranger station or park office before I go.
Then do these things:
Call ahead. Five minutes with a ranger beats an hour of chatbot answers. Ask about water sources and what time to turn around.
Pack for the bad version. Extra water, a headlamp, a paper map, a charged phone. If the bot gives you a water amount, treat it as the minimum.
Tell someone your turnaround time. If you’re not back by then, they call. Not you, them.
And never let an app or a chatbot tell you a wild mushroom is safe to eat. That’s a different kind of rescue.
📩 Send this to someone who leaves their phone in the car while hiking.
💬 YOU ASKED ME THIS
Image: ChatGPT/Kim Komando
“Kim, I love your AI newsletter. How can I tell if a video I see online is real or fake?” — Adam in Washington, D.C.
Great question, Adam. Short answer: Sometimes you can’t, but there’s a free way to check.
Google opened its AI detector to everyone. Go to synthid.com, sign in with a Google, OpenAI or Apple account and upload the photo, video or audio clip. It looks for Google’s invisible watermark, which is also built into content from AI tools made by OpenAI, Nvidia and Kakao.
🏷️ Check the tag
Think of the watermark like the tag inside a sweater. If the tag says “Made with AI,” you have your answer. But no tag doesn’t mean it’s handmade. Plenty of AI tools don’t use Google’s watermark, and watermarks can be stripped (people rip them off faster than mattress tags).
Here’s the twist: Even a hit won’t tell you whether the whole clip is fake or just touched up. And it can’t check pictures from Apple’s Image Playground yet.
Feel like fakes are everywhere? You’re not crazy. Google alone has watermarked more than 180 billion images and videos since 2023, and that’s only its own tools.
My take? A hit is a red flag. A miss means “still unknown.” Run the check before you share that wild video. When in doubt, don’t forward.
I washed my camera’s memory card. Now all my photos are watermarked.
📻 MY NATIONAL RADIO & YOUTUBE SHOW
📺 Flock cameras, flat butts & fiery robot retirements
A robot company couldn’t sell its old humanoids (too many secret parts), so it marched them into molten steel on camera. Terminator 2 called. That’s some severance package.
This week: The Flock camera fight from both sides, a $100 billion ad company accused of slipping vape ads into kids’ games, Meta’s Muse reading texts and bank accounts of 4 million people a week, a $2,000 folding iPhone and a $2 water bottle that cost one woman $2,400.
Plus, a license plate holder that fogs itself when Flock is watching, an AI morning show cohost that sent ratings up 75% and Amazon telling a shopper she has a flat butt. (Rude.)
Hit play below, so you know who’s reading your plate, what’s quietly billing you and where to buy dry ice. 👇
🎧 Or search “Komando” wherever you get your podcasts. I’m everywhere.
😏 REAL OR AI?
Image: @onapittance via X
Be honest: Would this place get your Saturday? Glowing gold lotus, candlelight, five kinds of massage, weird squiggly text and a woman who looks like she hasn’t opened her email since 2019.
Then again, everything before 2010 feels fake. Maybe because it’s all just pre-10s.
🥁 Drumroll please. Technically? Real. The image comes from a photo of an AI-made sign outside a spa. Check it out.
Somewhere, a graphic designer just felt a blood pressure spike.
BROUGHT TO YOU BY
🔥 The one thing worth saving first
4.7 ⭐ 8,800+ reviews
AI can back up your photos. It can’t replace your passport, your birth certificate or your Social Security card. One house fire, one burst pipe, and those are gone. The kind of gone you can’t undo.
This fireproof document bag can withstand up to 2,000°F and shrugs off water, too. Legal-size, two pockets, heavy-duty zipper. Fits everything that matters: certificates, insurance policies, passports, cash, the works. Toss it on a shelf, and sleep better tonight.*
Over 2,000 people bought this in the past month alone. Once you know it exists, you’ll wonder how you didn’t have one already. I looked for the catch twice. There isn’t one.
✅ Grab the fireproof document bag for 40% off today, just $18. → You’ll be so glad you did !
🫠 WTF (WHAT THE FUTURE)?
Image: @platcornweb via X
🪩 Hit me, bot, one more time
Colombian Kick streamer WestCOL brought a humanoid robot on air to show off its dance moves, and to be fair, it had some moves. But you know what they say, it’s all fun and games until a robot kicks in your display cabinet.
The 5-foot-7 machine, which reportedly cost over $100,000, whacked him in the back of the head, then kicked a glass display case into pieces before toppling over.
Check out its dance moves here. It’s the Terminator trying to win Dancing With the Stars after three Red Bulls.
🎬 END OF PROMPT
Here’s your haul for today. You learned how to turn your pet into a pumpkin masterpiece, let ChatGPT settle the fridge repair debate, avoid dangerous AI hiking advice, spot AI-generated fakes and watch a dancing robot destroy a display case. Not bad for five minutes of reading.
💦 The answer: D) About five drops. Google’s researchers found the median Gemini text prompt uses 0.26 milliliters of water and as much energy as watching nine seconds of TV. ChatGPT is in the same ballpark. OpenAI’s Sam Altman says an average ChatGPT query uses about one-fifteenth of a teaspoon of water.
So where did the bottle come from? In September 2024, The Washington Post teamed up with UC Riverside researcher Shaolei Ren and reported that writing a 100-word email with GPT-4 used about 519 milliliters of water. That’s a little more than a standard bottle. The headline went everywhere.
Here’s the catch. It’s math. The two numbers aren’t measuring the same thing. The bottle figure adds up the water used to cool the data center plus the water that evaporates at power plants making the electricity. Google’s five drops only counts what’s used on site. Why did the bottle number and the five-drop number get into a fight? They couldn’t see eye to eye on what counts as a fair pour. 💧 (Ouch. That hurt even me.)
📞 Before you go, that wasn’t your bank. It was a crook with AI. The voice sounds real. The caller ID looks real. Neither is. NordVPN warns you about risky calls and texts before you answer, and it never reads your messages. I use it on my own devices. Get 76% off. That’s $3.37 a month plus four extra months.*
💛 Look at you, all AI smart, taking on the day with the confidence of a 4-year-old in a Superman cape. Not a bad result from reading my free newsletter. Forward this to ONE person who needs to be with the AI program. Thanks for diving in. See you next time! — Kim
Kim Komando • Komando.com • 510+ radio stations • Trusted by millions daily
Photo credit(s): ChatGPT/Kim Komando, @onapittance via X, Andyer, @platcornweb via X
Companies and products denoted by an asterisk (*) within this publication are paid sponsors or advertisements. As an Amazon Associate, the publisher earns from qualifying purchases. Statements regarding products denoted by a double asterisk (**) have not been evaluated by the Food and Drug Administration; such products are not intended to diagnose, treat, cure, or prevent any disease. This newsletter is provided for informational and entertainment purposes only and does not constitute legal, financial, medical, or professional advice of any kind. Readers should consult with a qualified professional before making any decisions based on this content. The publisher disclaims all liability for any loss, damage, or injury resulting from the use of or reliance on the information contained herein.