• Splash of AI
  • Posts
  • šŸ’§ I showed AI my landscaper’s bill

šŸ’§ I showed AI my landscaper’s bill

He’s now adding plants at his own cost. Plus: ChatGPT is serving ads against your secrets, AI tool of the week and prompts you’ll use time and time again

šŸ™‹ā€ā™€ļø Hello to you this Thursday, friend. You’ve spent three years in law school. You’ve survived finals, moot court and enough caffeine to fuel a small city. Then you sit for the bar exam. One of the hardest tests in America. Only about half of first-time takers pass. GPT-4 took the bar exam. Not as a stunt. OpenAI wanted to know if it could reason through legal problems the way lawyers do. 

šŸ“š How did GPT-4 score on the Uniform Bar Exam? A) It failed. Didn’t come close, B) It scraped by with a barely passing score, C) It scored around the 90th percentile of human test takers or D) It got a perfect score. You’ll find the answer at the end of this newsletter, resting its case.

āœ… Quick favor: If this landed in your spam folder, your email has terrible judgment. Drag it to your inbox, and add me to your contacts. You really don’t want to miss next week. Let’s do this. — Kim

šŸ“¬ Someone forwarded this to you? Smart friend. Every Thursday, I cut through the AI noise and tell you exactly what matters, what to try and what to avoid. Sign up here.

šŸŽÆ YOUR AI POWER MOVE

Image: Gemini

This week: Your AI estimate assassin

Here’s how the game works. A mechanic, contractor, landscaper, plumber or hospital billing department writes you an estimate. It looks official. It has line items and totals. You have no idea if any of it is fair. 

They know that. So they charge what they think they can get. 

ā™Ÿļø Ruin their bold strategy

My landscaper quoted me $900 for eight cactuses. Something felt off. I couldn’t have told you exactly what or why. Just that feeling in your gut that says: nope.

šŸ—£ļø So I opened ChatGPT voice mode, turned on my camera and walked down the row. Three minutes later, I had a full breakdown by cactus. Turns out there were $20 plants in that mix. Not rare specimens. Not hard-to-source varieties. I never would have known that without the camera. I showed my landscaper what I found. He’s adding more plants at his own cost. Prickly, but point taken.

Here’s precisely how I did it:

  1. Open ChatGPT on your phone and tap the headphone icon in the bottom right corner to launch voice mode.

  2. Once you’re in the voice conversation, tap the camera icon to turn on your live camera feed.

  3. Point your camera at whatever you want identified. Hold your hand next to it, or set something familiar nearby, so the AI can judge the size accurately.

  4. Ask out loud: ā€œCan you identify what this is and give me a typical retail price range?ā€ Work through each item one by one.

  5. At the end, ask: ā€œBased on everything I showed you, what would a fair total price be?ā€

That’s it. You, your phone and a very informed second opinion.

šŸŽÆ Use this prompt

The best negotiating tool isn’t confidence. It’s homework. For estimates without physical items to inspect, open ChatGPT, Claude, Gemini or Grok and paste this:

Be a consumer advocate with deep expertise in [name expertise]. I received this estimate: [paste the full estimate here]. (1) Flag any line items that seem overpriced compared to typical market rates in [my city], (2) Identify any unnecessary charges for the job or ones that could reasonably be deferred, (3) Tell me the one question this provider is hoping I never think to ask.

That third question is the one. Every time.

Before you think this only applies to landscaping: Medical billing errors appear in an estimated 80% of hospital invoices, according to Medical Billing Advocates of America. The average overcharge runs $1,300 per patient. Yikes.

šŸ”§ Your move this week

  1. Find any estimate you have sitting around. Contractor quote, dentist bill, hospital invoice. Old ones count, too. Even your water bill that needs a ā€œget well soonā€ card.

  2. If it involves physical items, plants, parts or products, use ChatGPT voice mode with your camera. Hold something next to it for scale. Let it identify what you’re looking at.

  3. Paste written estimates into the prompt above. Walk back in with the questions. You’re not being difficult. You’re being informed.

Done, smarty pants.

Facebook | LinkedIn | X

 

Know someone who’s sitting on a repair estimate and doesn’t know if they’re getting ripped off? Forward this right now. You might save them hundreds before the weekend.

šŸ’¼ YOUR AI EDGE

Your rƩsumƩ is being buried alive

Here is what’s happening when you apply for a job and hear nothing back. Nobody can find you.

98% of Fortune 500 companies use Applicant Tracking Systems. That software collects and organizes every application. The average corporate job posting attracts more than 250 applicants. 

ā±ļø Count down

The problem isn’t that a robot rejected you. When a recruiter searches the system, they use keywords pulled straight from the job description. If your rĆ©sumĆ© doesn’t echo that exact language, you don’t surface.

You’re in the system. You’re just invisible. Like a sock in the dryer. Gone. AI fixes this in 60 seconds. 

šŸŽÆ Use this prompt

Open ChatGPT, Claude, Gemini or Grok and paste this:

You are a professional rĆ©sumĆ© strategist. Here is my current rĆ©sumĆ©: [paste your full rĆ©sumĆ© text]. Here is the job description I’m applying for: [paste the full job description]. Please: (1) Identify the exact keywords and phrases from the job description that are missing from my rĆ©sumĆ©, (2) Rewrite my top three experience bullet points to mirror the language of this specific posting, (3) Rewrite my professional summary to speak directly to this role, (4) Tell me the single biggest reason a recruiter skimming this would move on to the next candidate.

You’re not gaming the system. You’re speaking the same language as the person trying to find you. Like the Coast Guard recruiter who asked a man, ā€œDo you know how to swim?ā€ The man replied, ā€œWhy? Have you run out of ships?ā€ (I thought maybe you needed a smile after all that.)

Easy.

šŸ“» MY DAILY RADIO FEATURE & PODCAST

šŸŽ™ļø I asked AI what it knows about me. Big mistake.

My shows hit 500+ stations and every major podcast platform. I asked an AI what it knows about me. I’m telling you right now: Try it before you listen to this episode. Also on deck: the jobs that are safe from AI, Iran-linked hackers target Stryker medical devices and the AI-generated actress Hollywood is losing sleep over.

šŸŽ§ Listen now.

šŸ’° AI DEAL OF THE WEEK

šŸ’» Your AI sidekick: Acer Aspire 16 Copilot+ PC ($545)

Image: Acer

You open your laptop to ā€œquickly check one thing.ā€ Forty-five minutes later, you’ve got 12 tabs open, half a sentence written and no idea where you started. Sound familiar?

This Acer Aspire 16 is here to clean up that mess. It’s got Copilot+ PC, which means built-in AI helps you get stuff done. Ask it to summarize a long doc, rewrite an email or find that one file you forgot to name properly. It’s like having a very fast, very patient assistant who never needs coffee.

The 16-inch touchscreen is big, bright and buttery-smooth at 120Hz. Translation: Scrolling feels slick, and everything looks sharp. Inside, there’s 16 GB of memory, a fast 512 GB SSD and Wi-Fi 7, so downloads don’t crawl like it’s 2009.

āœ… Check out this solid budget AI laptop now. It comes with a 30-day return policy.

āš ļø THE AI TRAP

People tell ChatGPT things they don’t tell their doctor. Their real financial situation. The symptom they’re scared to google. The marriage problem they haven’t said out loud to anyone.

šŸ“¢ Ad-vice you didn’t ask for

OpenAI started running ads inside ChatGPT in February. They use the topic of your conversation, past chats and previous behavior to decide which ad you see. Ask about knee pain? Expect a sponsored result. Stress about money? Same.

OpenAI says ads won’t influence answers. Facebook said the same thing about our Feeds. Here’s what you need to know.

šŸ”’ Your options

  • Free tier, keep the ads: Do nothing. Ads appear below responses, labeled ā€œSponsored.ā€

  • Free tier, ditch the ads: Go to Settings > Ads Controls > Change plan to go ad-free > Reduce Message Limits. Ads disappear. You’ll get fewer daily messages. No cost, just a trade-off.

  • Opt out of personalized ads: Go to Settings > Ads Controls and toggle off Personalize ads. Ads become generic instead of targeted to your conversations.

  • Go completely ad-free: Upgrade to ChatGPT Plus at $20/month. No ads, period.

  • Switch tools: Claude, Gemini and Grok are currently ad-free. Worth knowing.

IN PARTNERSHIP WITH

Real results. Real power.

I’m always searching for ways to help you stay sharp, strong and feeling your best. NativePath Collagen is a game-changer for joints, skin and confidence.** 

Bonnie says:

ā€œI’m 73 and someone thought I was in my 40’s. I’ve been taking NativePath for a few months. My knees used to ache, but this helps a lot. I really notice when I skip it. Two scoops in my morning coffee. It’s made a huge difference in my skin, hair and nails. I never leave reviews, but this one is a must.ā€

NativePath Collagen helps support your body, the engine behind it all.

šŸ‘‰ Get free shipping and free gifts today with NativePath! →

These statements have not been evaluated by the Food and Drug Administration. This product is not intended to diagnose, treat, cure, or prevent any disease. Individual results may vary.

🤯 ā€œI HAD NO IDEAā€

You may have seen the story about people getting paid $20/hour to do their dishes on camera for AI to learn how to move its hands. (If not, the full breakdown is here at Komando.com. Wild read.) But there’s a whole other tier most people don’t know exists, and it pays a lot better than your sponge ever will.

🧠 They need human brains 

AI companies need people with real professional expertise to train AI how to think in high-stakes situations. How to interpret a lab result. How to read a contract. How to spot a structural flaw in an engineering plan.

Outlier.ai connects over 700,000 MAs, PhDs and college grads with AI companies that need expert human feedback. Economists earn up to $55 an hour. Coders, writers, lawyers and STEM specialists land anywhere from $15 to $60+. No set schedule. No minimum hours. Work between meetings or after the kids go to bed.

FYI, reviews are mixed on work consistency. Treat it as supplemental income, not a full-time gig. 

šŸŽÆ Use this prompt

Open ChatGPT, Claude, Gemini or Grok and paste this:

You are a career strategist specializing in the AI training economy. Based on my professional background below, identify which AI companies are most likely to pay for my expertise, what specific skills they’re recruiting for right now and exactly how I should position my experience in an application. Here’s my background: [paste your rĆ©sumĆ© or a quick summary of your work history and credentials.]

Two things will happen. First, you’ll find out which tier you qualify for. Second, the AI will write you a pitch that sounds like someone who knows this industry. Because, well, it does.

šŸ”§ Your move this week

  1. Go to Outlier.ai and browse openings. Takes five minutes to see if your credentials fit a project.

  2. Paste that prompt into any AI tool with a quick summary of your background. Let it match you to the right opportunity.

  3. Apply. The onboarding takes 30 to 90 minutes, but once you’re in, the work comes to you.

The robots are learning to think either way. The only question is whether your expertise ends up as free training data or extra money in your paycheck.

šŸ› ļø YOUR TOOL OF THE WEEK

Suno

WHAT IT DOES

Turns a text description into a fully produced original song, complete with real-sounding vocals, instruments and mixing. In about 30 seconds.

WHO IT’S FOR

Any of you who ever wanted to make a song and assumed you couldn’t because you’re not musical. You don’t need to be.

PRICE

Free. 10 songs per day. Pro plan $8/mo for commercial rights and higher limits.

TRY IT
MY VERDICT

I typed ā€œupbeat country song about a golden retriever who steals socks and has zero remorse,ā€ and 30 seconds later, I had a fully produced track with real vocals and a banjo solo. Bella and Abby listened to it once and left the room. (Honestly, fair.) But here’s where it gets powerful: My brother-in-law used Suno to write a personalized anniversary song. Referenced their first date, their dog’s name, the diner where he proposed. My sister sobbed. He got all the credit. The AI did the work in 90 seconds. Use it for a birthday song that mentions someone by name. A silly graduation anthem. A lullaby with your baby’s name in the chorus. A memorial tribute with the exact stories you want the world to remember. This is one of those tools you try once and immediately text five people about. Zero musical talent required. I mean that. None whatsoever. Free to start. Go.

Neat.

🫠 WTF (WHAT THE FUTURE)?

Image: Walt Disney Imagineering via YouTube

ā„ļø Do you want to build a robot?

Disney made Olaf real, and somehow he has better posture than most of us. 

Meet Walt Disney’s half-secret Imagineering facility. Their fully self-walking Olaf (no tracks, no visible controls) is an AI-powered waddling bundle of joy about to be unleashed in Hong Kong and Paris. Hope it doesn’t have a meltdown.

Behind the scenes, they’re teaming up with NVIDIA, Epic Games and Meta to build characters that move, react and exist like they’ve escaped the movie. Think Spider-Man doing flips, Mandalorians roaming, the whole cinematic universe leaking into reality at a Disney Park near you.

We’re all probably one bad software update away from taking orders from an army of Minions. Maybe they’ll Gru on us. Watch more here.

šŸ—£ļø Text / Post This Stat

AI companies are paying people $20/hour of home chore footage to train humanoid robots. Scale AI has already collected 100,000 hours of it. Goldman Sachs says the humanoid robot market could hit $38 billion by 2035. Your dishes were always worth something. Now there’s a check for it. Get the weekly playbook free at SplashOfAI.com

šŸŽ¤ SIGNING OFF

šŸ‘ØšŸ»ā€āš–ļø The answer is C) It scored around the 90th percentile of human test takers.

GPT-4 scored approximately 298 out of 400 on the Uniform Bar Exam. That’s not only passing. That’s beating 9 out of 10 humans who took the same test. For comparison, GPT-3.5 (the version before it) scored around the bottom 10th percentile. One generation of AI later, it leapfrogged almost every human in the room. The same model also passed the LSAT, the SAT and multiple medical licensing exams. It didn’t study. It didn’t stress. It knew.

šŸŽø One for the road: My friend who absolutely loves U2 passed the Bar Exam. He told me everything he does now will be Pro Bono. (Was that a groan I heard?)

Phew, that was a lot to take in. Give yourself a pat on the back. Until next issue, the future isn’t coming for you. You’re coming for it. šŸš€ — Kim

Know someone who thinks AI isn’t going to touch their life? Forward this whole issue. By the time they finish reading, they’ll know it already is. And they might make $80 doing dishes this weekend.

Kim Komando • Komando.com ā€¢ 510+ radio stations • Trusted by millions daily

HOW’D I DO?

What did you think of today’s issue?

Photo credit(s): Gemini, Acer, Walt Disney Imagineering via YouTube

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.