When Google Gemini replaced Google Assistant on a reader’s Android phone, her favorite morning routine disappeared.

“Until Wednesday all I had to do was say ‘OK Google’ and it would give me a good morning, tell me the weather and the time and a joke, and read me the news,” Sheila said. “A pox on them all for changing my routines.” Almost as bad, Alexa changed its voice.

“Am I too old for all this?” she wondered. “After all, I learned to talk by listening to Franklin Deleano Roosevelt’s fireside chats on an old-fashioned radio console.” No way. She followed my instructions perfectly.

To go back to Google Assistant on an Android phone, first open Settings, go to Apps, then Default apps, and look for the digital-assistant setting. Choose Google Assistant instead of Gemini. The exact wording may vary by phone.

If Alexa Plus AI has replaced the original Alexa, try saying, “Alexa, disable Alexa Plus.” If only the voice has changed, say “Alexa, change your voice.”  It will give you several options.

AI sometimes bugs me too. For example, I ditched Microsoft Edge when every new tab started opening up with Copilot AI. I hadn’t asked for an AI chat, I just wanted my Gmail or a Google search.To get rid of it, I clicked the three dots in the upper right of my computer screen. Then I clicked “Settings.” Next, I clicked “Copilot and AI.” Finally, I toggled off the “Copilot new tab page.”

ChatGPT Joins Microsoft Word

When editing a long memoir, I discovered four nearly-identical passages. They all repeated earlier anecdotes, with slight variations. What a mess. So if it ever happens again, I might use the new ChatGPT option in Microsoft Word to sort it out  rather than copy and paste words from a document to ChatGPT.

ChatGPT is now a free add-in for Word, Excel and PowerPoint. Once you’ve added it, you’ll see ChatGPT’s white-and-gray icon in the upper right of your screen in all of those programs. In Word, it could point out inaccuracies, omissions, needless words and more. Inside Excel, it could help you build a spreadsheet or figure out a formula. In PowerPoint, it could polish your slides. But be aware that any document content, chats or attachments could be shared with Microsoft and OpenAI, so that ChatGPT can read and update your files. It’s not meant to spy on you.

To set it up, start by Googling “Microsoft Marketplace.” Once you’re there, search for “ChatGPT,” click “get it now” and follow the prompts. Inside Word, Excel and PowerPoint, click the “Home” button. Now look to the far right and click “Add-Ins.” Then click “ChatGPT.” A ChatGPT button will add itself.

Wearables Get a New Brain

At the Snapdragon Summit last week, PrismML demonstrated “Bonsai,” a new technology that squeezes powerful AI into far less computer memory. As the crowd oohed and aahed, a Bonsai AI system ran entirely on a pair of smart glasses powered by a Qualcomm chip, instead of sending every question to a giant data center.

This is part of a growing trend called on-device AI. Instead of relying entirely on the cloud, some of the AI runs right on your phone, glasses or other gadget. That can make responses faster while keeping more of your information on the device. Some on-device AI can even work without an internet connection.

You may also hear about “small language models,” or SLMs. They’re compact versions of the enormous AI systems that require much more computing power. Bonsai takes the idea further by drastically reducing the amount of memory an AI model needs.

Maxton Hall

Dub-a-Dub Dub

On Amazon Prime Video, I just watched a half hour of the hit German series, Maxton Hall. I wanted to see AI dubbing at work.

In ordinary dubbing, the actors’ lips don’t match what they are saying. Under the new system, human voice actors record the English dub first. Then the AI and visual-effects software alter the actors’ lip movements to match what the actors just said.

When I watched Maxton Hall, the actors’ lips mostly looked in sync.Only in close-ups did I notice imperfections.

This isn’t the first use of AI dubbing. TrueSync technology was famously used in the 2022 thriller “Fall,” where it altered the actors’ lip movements after more than 30 cuss words were replaced with PG-13-friendly dialogue.

Robot News

A new, child-sized robot called the PrimeBOT Q1 is capturing hearts all over. It recently went on sale in China for about $2800.  I remember when an ordinary laptop cost  more than that.

At 2.9 feet tall and 33 pounds, the PrimeBOT can hold your hand and walk around with you. Designed to be your companion, it can dance, talk to you and perform more than 10 martial arts movements. Change its appearance with swappable 3D-printed panels, voice choices, personality settings and facial expressions. Its programming tools are open to developers, which should encourage robot owners and educators alike to create new skills, motions and companion features. You can sign up for more information at www.primebot.cn.

Internut

Search on “Same Wound, Two Ages: The Conviction Crisis and the Editing Crisis” for a podcast about a condition shared by teenagers and business leaders alike: the pressure to change yourself to fit in.

Filed under: AI, Internuts, robots, wearables | Add New Comment »

My late husband Bob tried to get hyperbaric oxygen therapy for his diabetic foot ulcers, but he couldn’t get an appointment. The only center near us reserved its chambers for emergencies such as carbon-monoxide poisoning, decompression sickness, gas embolism, severe crush injuries and certain kinds of infections. 

But an unusual technology is coming to the rescue for these stubborn sores. It uses fish skin as a wound dressing. Sound crazy? It isn’t.  

A company called Kerecis starts by taking skin from North Atlantic cod, removing the fish cells, sterilizing the skin and freeze-drying it, while leaving the skin’s microscopic collagen framework largely intact. Doctors rehydrate the sheet and place it over or into a wound. This serves as a temporary scaffold. The blood vessels and the patient’s own cells grow into the matrix as the fish material is gradually absorbed by the body. 

Kerecis’s wound dressing, called “MariGen,” is classified by the Food and Drug Administration as a medical device. It was first cleared in 2013, but now we have more evidence that it works. A recent randomized trial published in NEJM Evidence studied 255 patients with deep diabetic foot ulcers, some extending to bone or tendon. At 16 weeks, 44% of patients receiving intact fish-skin grafts had healed wounds, compared with 26% receiving standard wound care. At 24 weeks it was 55% versus 38%. Two caveats: Karecis helped fund this study, and the technology  shouldn’t be used in people with fish allergies.

According to Kerecis, insurance coverage expanded this year by more than 40 million insured people, including additional private insurers and state Medicaid programs. Medicare has a billing code for Kerecis fish-skin products, and its revamped 2026 payment system separately reimburses qualifying skin substitutes. 

Data Center Mania

A reader wrote to say he’s upset about big data center hogging of water and electricity. Also, in his neighborhood, he says, the secrecy involved in expanding these centers could mean that taxpayers are on the hook for some of the cost.

What if we pair data centers with nuclear power? That could add huge amounts of around-the-clock electricity. In addition, water use could be cut sharply, thanks to newer data-center cooling systems. But of course, nuclear power still requires lots of cooling water, and many of the biggest tech-nuclear deals still depend on regional electric grids. So the nuclear plan isn’t perfect. 

Four nuclear developers have attracted some big tech partners. Meta has agreements with TerraPower, founded by Bill Gates, and Oklo. Amazon has invested $500 million in X-energy. Google has a nuclear-energy deal with Kairos Power. 

Already, data centers and nuclear plants are getting paired. Amazon bought a 960-megawatt data-center campus beside Talen’s Susquehanna nuclear plant in 2024. The power agreement has since expanded to as much as 1,920 megawatts through 2042.  Also promising is Constellation Energy, the largest commercial nuclear operator in the U.S. It signed major multi-decade power purchase deals with Microsoft and Meta to support their growing data-center power needs. 

New Key to Alzheimer’s

Thanks to “ShortStop,” an AI tool developed by the Salk Institute, researchers have found a hidden world of more than 1,000 microproteins that offers a new perspective on Alzheimer’s disease.

One newly discovered microprotein was unusually scarce in Alzheimer’s brains. When scientists removed it from brain immune cells in the lab, the cells had trouble producing energy. That gives researchers a new clue. If further research confirms its importance, scientists might someday develop a drug that boosts its activity or mimics what it does. 

Siri Gets AI

With iOS 27, an AI-equipped Siri is rolling out in beta on iPhone 15 Pro or 15 Pro Max, or any iPhone 16 or later. She can now draw on information from both your phone and the web to answer questions.  

If you ask her for a friend’s address, or some other tidbit, she’ll produce it pronto. She can also answer more complex questions. But in my tests, she usually enlisted ChatGPT’s help after I said OK. For example, this happened when I asked whether it was better to buy books from the local bookstore or Amazon. Sometimes, however, she couldn’t answer at all. Also, when I asked her to remind me to go to the library in five minutes, she put the reminder on the screen instantly but then forgot to remind me.

In other iOS 27 news, your ringtone no longer has to have the same volume as your alarm. To change it, go to “Settings,” then “Sound & Haptics,” and turn off “Match Ringtone Volume,”  under “Alarms and Timers.”  Your other alerts and system sounds don’t have to match your ringtone or alarm volume either.

Headset for Depression

For the first time, some Americans with moderate-to-severe depression can use an FDA-approved at-home headset for 30 minutes at a time to electrically stimulate a brain region involved in mood. The FL-100 device from Flow Neuroscience, which uses transcranial direct current stimulation, became available by prescription across the U.S. on Sept. 1.

A randomized, double-blind, sham-controlled trial was published in Nature Medicine. In the 173-person analysis, those receiving real stimulation improved significantly more after 10 weeks than those receiving sham stimulation. But the FDA called the benefit modest and it costs $2,200 for the initial 10-week treatment course, followed by $325 every four weeks for ongoing maintenance. It might be better to get a dog.

Filed under: AI, Apple, health | Add New Comment »

When an old high school friend reached out to me through Classmates.com, I had a fascinating discussion, thanks to input from Google Gemini.

First my friend Stephanie and I reminisced about the good old days on the high school newspaper staff. One time, some sang “Tea for Two,” while others pounded on their desks to the beat, pulled the blinds up and down rhythmically or used a ruler as a baton to add to the merry mayhem. Meanwhile, our mild-mannered teacher ignored everything. He wasn’t even that mad on the day when students set up the whole classroom on the street, complete with the American flag in its holder. Or when a few guys locked me in the dark room with the editor.

Now an artist, Stephanie offered me one of her 24 gorgeous paintings for free. “But which one should I choose?” I wondered. To find out, I opened her website in Chrome and clicked “Ask Gemini.” This allowed Gemini to examine the website I was looking at. I also uploaded photos of my living room and den. I wondered whether I should hang the painting above a 31-inch bookcase near a Renoir poster in the living room or on a narrower wall above another bookcase in the den.

But which painting was best? I asked Gemini to choose the two that most people would prefer. One happened to be my favorite. It also confirmed my instinct about the proper placement.

“That is a wonderful choice!” Gemini gushed. “The house with the reddish roof, cheerful front door, and green plants is warm, full of character, and wonderfully inviting. Its rich, earthy tones and vibrant foliage will look fantastic above your 31-inch bookcase, tying together the warm wood furniture and picking up nicely on the subtle red accents in your room. Enjoy your new piece of art!” “I’m honored to be next to the Renoir,” Stephanie said.

Not Listening

Did you ever wish you could hit rewind instead of asking, “Honey, what’s that you were saying?” Soon you can. “Live Rewind,” coming in beta later this year to the Apple Watch Series 12 and Apple Watch Ultra 4, will let you view a text transcription of the last 15 seconds of whatever conversation is going on around you. Just double press the Digital Crown to instantly pull up that missed bit of dialogue.

Fortunately, it doesn’t make or store an audio recording. The sound is processed in a protected buffer and immediately deleted. Nor can you use Live Rewind secretly: when you trigger it, the watch sounds a chime and displays a microphone indicator. But some still say it’s creepy.

Live Rewind requires an Apple Intelligence-capable iPhone 16 or later, but won’t work with the budget phone, the 16e.

ChatGPT Solves Printer Snafu

I was trying to print 260 pages in Korean, but my printer kept stopping every few pages with a “paper out” message, no matter how I loaded the rear paper feeder. Yet it labored on, hacking like a chain smoker every once in a while.

Finally, it got stuck on “paper out.” As its power light blinked ominously, Gemini suggested I clean the rubber rollers. “Should I use a screwdriver to open it up?” I wondered.

That sounded daunting. So I asked ChatGPT. It instructed me to shine a flashlight into the belly of the beast and upload a photo of what I saw. Aha! A pencil had gotten stuck down there, almost perfectly intact. Only an inch-long smidgeon had chipped off.

Unfortunately, the wayward pencil seemed to have damaged the printer’s pick-up mechanism, ChatGPT said. But instead of paying for an expensive repair, I bought the Epson XP-4200 for $60 from Target. It prints beautifully, although Epson’s replacement cartridges are expensive– $23 for the black cartridge alone. My old printer is going to Goodwill with an explanatory note and four nearly-full bottles of ink.

Google Spies

After I told a friend that I often let Google Gemini examine the websites I’m looking at, she said: “Don’t you worry that Google spies on you?”

“I understand why other people do,” I said, “but I don’t mind getting ads tailored to my interests. I don’t think it’s worth Google’s while to spy for other reasons.”

Gemini said that Google isn’t manually monitoring me as an individual or spying in the traditional sense. “Most data processing is automated by algorithms designed to detect patterns across hundreds of millions of users, not build a targeted docket on a single person for fun.”

But it also pointed out three things I’d missed. First, sharing a browser tab with Gemini can inadvertently expose private messages, medical records or emails from people who never consented to have their data analyzed. That’s unfair to them.

Second, though innocent as a newborn lamb, I could be the victim of a warrantless search. Government agencies are generally supposed to get permission before seeking private information, but can buy commercially available data instead, Gemini pointed out.

Third, I could lose money. If websites tap my data to infer my income bracket or how badly I want something, they might show me higher prices. The Federal Trade Commission says that companies can use things like location, browser history, demographics, shopping history and even mouse movements to customize prices, discounts or the products consumers see.

Filed under: AI, Apple, art, Google, wearables | Comments Off on ART AND AI

I love the debates at OpentoDebate.org. But if I’m short on time, I use Google Gemini or ChatGPT to summarize the exchange and tell me which side made the stronger case.

In one titled “Is America Ready for a Four-Day Work Week?” the majority sided with economist Juliet Schor. That’s because she presented stronger, data-backed evidence from numerous corporate pilot programs that showed immediate benefits to both employee well-being and corporate productivity, Gemini said. She also said that the four-day work week will probably evolve naturally before it’s mandated by law. Her rival, journalist Joanne Lipman, pointed out that many executives, especially in the artificial intelligence community, claim they want a shorter work week but constantly push for longer hours and fewer remote jobs.

In “Open to Debate,” both the live audience and the online audience get to vote on the “pro” or “con” side of a question before the debate occurs and afterwards. Whether online or in person, however, you must vote within a designated time limit.

 In a recent debate titled “Can Humans Adapt to Climate Change?” 61% initially voted yes, 23% voted no, and 16% were undecided. After the debate, 50% voted yes, while the “no” vote rose to 42%.

Fixing a Fuzzy Photo

A friend took a great photo of me with colleagues visiting from Italy. But the photo’s resolution is only about 1% of normal. So it looks extremely fuzzy when viewed on anything larger than a phone.

To sharpen it, I first tried Use.AI, which uses AI tools to guess the details it cannot see. No wonder it gave my face a mean expression. Meanwhile, ChatGPT gave me extremely almond-shaped eyes. Upscayle.org got stuck in the upscaling. GIMP, a free Photoshop alternative, kept things fuzzy even after quadrupling the resolution. Finally, my friend’s boyfriend sent the uncompressed versions from his phone.

To make sure an iPhone sends the original photo, open Photos, select the photo, tap the “Share” icon, then “Options” and turn on “All Photos Data.” You can then send it by AirDrop to a nearby Apple device or create an iCloud link to send it to someone farther away.

On an Android phone using Google Messages, select the photo, tap HD or HD+ and choose “HD+ Original quality.” Alternatively, look for your phone’s DCIM/Camera folder and share photos directly from there. 

AI Tools for Neurological Diseases

“What AI tools are used to diagnose rare neurological diseases?” a reader asked. So I asked AI.

If the reader is seeking an answer for personal reasons, Gemini said, he should go to a neurogeneticist or rare-disease clinic that uses AI. Tools include Exomiser and Face2Gene, along with other programs that use a patient’s symptoms to prioritize possible genes and diseases. AI can also find out if there’s a clinic near you. 

AI Skips to the Best Parts

Using the Kindle app on my phone, I’m reading the great novel “Once an Eagle,” by Anton Myrer. This year, it was named the Commandant of the Marine Corps’ “Commandant’s Choice.” It has long been used in professional military courses. 

I’m finding it a real page-turner, full of action and even romance, with insights into World War I, World War II and the Vietnam War. But because it runs to 1,312 pages in the Harper Collins edition, I asked ChatGPT or “Chatty,” as a friend calls her, to list its most important pages, anti-war insights, leadership quotes and juicy scenes, just to be sure I don’t forget to talk about them at an upcoming book discussion.

In response, Chatty gave me a great list of themes, episodes and leadership quotes worth talking about. It also offered this tidbit: “The fascinating contradiction is that ‘Once an Eagle’ is simultaneously one of America’s most famous novels about military leadership and a deeply anti-war novel.”

Super Flosser

Ninety percent of us admit we don’t floss as often as we should, according to Dyson, the vacuum maker. So it came out with “CameraJet,” a $499 high-tech toothbrush.

The CameraJet uses a tiny camera to spot the flossable areas. Dyson says its precision-flossing helps prevent plaque from building up.

Here’s how it works. In the handle, there’s a 100,000-pixel macro lens camera that analyzes 28 images a second. An AI algorithm in the app identifies the gaps between your teeth before triggering the pump. 

I thought this meant a mighty blast, but the toothbrush handle holds less than a tablespoon of mouthrinse (12.5 milliliters) in its tank. A precision burst fires about 0.15 milliliters, or five thousandths of an ounce. Incredibly, because the force is great, you will still feel a pop of pressure as it cleans between your teeth.

A PC Magazine reviewer said he thought it was a gimmick until he saw it in action. Now he’s going to get one. Unfortunately, it went out of stock at Dyson.com just five days after it was unveiled at IFA Berlin, one of the world’s largest technology shows. 

Internut

Search on the phrase “Scientists Played Bach to Sleeping Newborns and Discovered They Can Track Rhythm Just 48 Hours After Birth” to uncover an amazing story published in PLOS Biology. Just two days after being born, the babies’ brains could already predict surprisingly complex rhythms, but not melodies. 

Filed under: AI, health, Internuts, photos | Comments Off on AI FOR DEBATES

I didn’t expect Google Gemini to be business savvy. But it taught me how to get my money back from an unscrupulous chiropractor.

I was roped into paying over $3,400 for back adjustments I didn’t need and foot therapy I did. Somehow the guy convinced me that an invisible malady would overtake my back if I didn’t endure 24 sessions.

Although I was fooled by the X-rays of my spine, Gemini wasn’t. It said that my variation from the norm was so slight that the way I was standing might have contributed to it. Besides, I have zero back pain and no stiffness. “The spine adjustments sound like a scam, and nothing to do with your problems. They are just trying to milk you,” said New York Times best-selling author and board-certified physician Joel Fuhrman, M.D., after I showed him the X-rays.

Next, I texted the chiropractor’s office to tell them I was canceling but got no response. So I called them up. The receptionist said someone would call me back in five minutes, but no one did. Gemini encouraged me to call back in late afternoon and gave me a script. That did it. They promised a refund. But if the money doesn’t land in my account in seven business days, I will protest. Gemini said I could dispute the charge with my credit card company for services I hadn’t received.

ChatGPT health

When you go to ChatGPT.com or launch the app on your phone, look for the word “Health” in the menu. ChatGPT Health began rolling out to U.S. adults in July 2026. It allows you to connect your medical records and discuss them with ChatGPT. This discussion can also include fitness information from Apple Health on your iPhone, if you agree.

The long back-and-forth with “Chatty” (ChatGPT) led me to explore new options, like SoftWave TRT (Tissue Regeneration Therapy). Chatty also pointed out that I had a referral to a foot and ankle surgeon, which somehow I missed. Fortunately, I also learned that going to a surgeon doesn’t necessarily mean surgery. We even had a discussion about gym shoes.

AI finds witnesses

These days, police officers may not need license plates to find witnesses to a crime. Flock is testing an artificial intelligence system that can find them without that, according to Wired Magazine.

Thanks to Flock’s new AI software, OS Investigate, camera data can be combined with case files and other records to uncover the names, home addresses and relatives of potential witnesses. To start, an officer might say: “Find me witnesses based on the vehicles most seen in Central Park in the last two days during the hours of 9 p.m. to midnight.” But these features are still in the testing phase, so they may be left out of  the final product.

Meanwhile, 46 communities have canceled Flock contracts, 20 have rejected proposed contracts, 19 have deactivated Flock cameras, eight have removed cameras and 12 have paused or suspended programs, according to a group calling itself “DeFlock the USA.” But that still leaves 90,000 Flock cameras all over the U.S., as well as the Motorola equipment that works with Flock.

Photo courtesy of ars technica

Mosquito-Zapping Drone

A mosquito-zapping drone from Tornyol weighs less than a golf ball. Though only 1.4 ounces and still a prototype, it uses sonar to detect mosquitoes by their distinctive wingbeat, then whams them with its propellers.

Unfortunately, in the company’s demo, it targets a moth, which is much larger, slower and easier to catch. In addition, it has to sit on its charger 90.9% of the time just to make a three-to-five minute flight, according to calculations made by New Atlas.

There’s also a cost consideration. You either pay $1,100 or sign up for the $50-a-month subscription plan. But I love the basic idea and it’s impressive that it was backed by Y Combinator, a premier technology accelerator that invests $500,000 in early-stage companies.

Amazon Drone Deliveries

Look up in the sky. It’s a bird, it’s a plane, no, it’s an Amazon drone! Here comes your package.

Amazon’s delivery fleet will reach nearly 500 cities and towns by the end of this year, according to Digital Trends. That’s up from 11 currently.

The deliveries are free for Prime members on orders totaling $50 or more. If the order totals less than that, it’s $2.99 per delivery, and non-members pay $4.99. Packages can’t weigh more than five pounds, but more than 60 percent of Amazon’s most frequently purchased items qualify, including groceries, electronics, cosmetics, and medications.

There’s been some bad publicity, however. A drone struck a crane in Arizona in October 2025 and cut through an internet cable in Texas in November. Competitors include Google Wing, which has already made over one million deliveries by drone, and Zipline, with over two million across four continents.

Internut

“New sodium ion battery stores twice the energy and desalinates seawater.” Search on that phrase to read about a breakthrough. Scientists used to think the water in a battery causes damage. But scientists from the University of Surrey discovered that by leaving water in sodium vanadate hydrate, a cathode material, they could nearly double its charge capacity, speed charging and keep it stable for hundreds of cycles. The researchers also showed that the system could remove salt from water while operating.

Filed under: AI, Amazon, drones, health | Comments Off on GEMINI SAVES THE DAY

When my niece recently applied for a PhD in physical therapy, she needed ChatGPT’s help to consolidate a list of 20 varied work experiences. That’s when she ran up against a paywall.

It may seem like the instant you start uploading documents, creating images or analyzing data, the paywall appears. But ChatGPT now offers unlimited texts and chats if you keep it casual. If it does ask you to pay, just switch to another AI service, such as Google Gemini, Google AI Mode or Microsoft Copilot.

What about Claude? If your queries are complex, Claude AI may ask you to come back in five hours. For example, when I wanted to translate a professor’s memoir into Korean, I’d barely started working when it asked me to come back later. “What should I do now?” I asked Claude. (In general, if I have a question about AI, I ask AI.) It introduced me to “TypingMind,” which helped get the book translated for less than I would have paid.

It wasn’t the best translation. In a side-by-side comparison with the English text, I discovered that Claude skipped whole paragraphs sometimes. So I asked GPT (or “Chatty,” as a friend calls her), if it would rather start from scratch. No, Chatty said, it appreciated Claude’s translation. But it made improvements to the Korean style and grammar and filled in the omissions. I paid $20 for one month’s usage, then canceled.

Google Gemini’s capacity refreshes every five hours until you reach a weekly limit, but I’ve never reached that, despite what seems like lots of queries. Like Anthropic’s Claude and Microsoft Copilot, it has given me great personal advice, answers to compelling questions, travel advice and directions. For example, Gemini talked me through the gigantic O’Hare airport when I was coming off a flight one night. I had to meet my stepson in a distant terminal and he hadn’t told me what gate he’d be at, but Gemini figured it all out. I used the microphone in the Gemini phone app to ask my questions so I could hear the answers without having to look down at my phone.

I like Copilot mainly for shopping advice and recipes, though it can answer tough questions too. Copilot uses the same OpenAI models that power ChatGPT, but Microsoft adds its own personality, safety features and Bing web search.

Fitness Fiasco

My new Hume Health fitness tracker seemed great at first. But now I don’t trust it.

I found out that the previous version, the Hume 1.0, was virtually identical to a $23 Chinese knockoff, despite costing $333 more, according to Dr. Rob ter Horst, a postdoctoral scientist with over 409,000 subscribers on YouTube. In ter Horst’s comparison, the Hume’s tiny logo was the only thing missing on the cheapie version. The Hume app even recognized the clone as a genuine Hume band when connected via Bluetooth.

According to ter Horst, the original website used a testimonial from a fictitious author whose stock image appears in back pain and real estate ads. Though it’s no longer on the site, its prior existence makes me suspicious.

Moving on to the latest version, the $249 Hume 2.0, ter Horst says it’s the worst wearable he ever tested, and he’s tested a lot. For example, ter Horst’s blood pressure readings were too uniform to be believable. So were mine. I also noticed that my sleep score didn’t budge, even after a rocky night. Ter Horst used a top-of-the-line electroencephalograph for comparison purposes and found appalling differences. But even if you only care about fitness tracking, the Hume 2.0 disappoints. It currently tracks only running, walking and weightlifting and doesn’t do a good job of any, ter Horst says.

Alternatively, he suggests getting an Amazfit Helio Strap or a Fitbit Air for around $100, less than half the price of the Hume. In a chart ranking dozens of wearables, the Pixel Watch 5, Apple Watch Ultra 3, Oura Ring 5, and Whoop are in the top tier.

Obsessed with Health Wearables

“Why I’m obsessed with health wearables (and you should be too)” is a 13-minute talk at Ted.com from genome researcher Michael Snyder.

With a smartwatch and pulse oximeter, Dr. Snyder discovered he had Lyme disease when his blood oxygen dropped precipitously while his pulse rate went up. This info led to early treatment, avoiding complications.

Consequently, Snyder’s team at Stanford developed a smartwatch-based warning system that looks for changes in an individual’s normal resting heart rate. During COVID, the system generated advance alerts for 80% of infected participants, typically three days before symptoms appeared.

Snyder wears eight wearables, including four watches, two rings and a couple of hearing aids, though he looks young. I recently misplaced my Apple Watch and am presently enjoying an unwatched existence.

Robot News

–Soft flying robots are the latest thing in robotics. Like a flying pillow, they’re much less likely to hurt humans than the metallic kind. They’re also much quieter, create less disturbance in the air, and float upstairs easily. The whale version might be your next dance partner, buddy or floating alarm clock to nudge you awake.

–Robots are starting to wear cozy-looking bodysuits to blend in with the family. The Neo, from startup 1X, will acquire skills over time, besides looking nice. It’s expected to ship later this year.

Filed under: AI, health, robots | Comments Off on AI FOR FREE

When a reader clicked the “Ask Gemini” button in Google Chrome and asked a question, it froze before giving him an answer. “Was it because I was using the free version?” he wondered. Nope.

When Gemini freezes, it’s a bug, not a paywall. It could happen when you ask unusually long or multi-part questions. To get around it, ask again and this time, keep your question short. But if you want, you can upload up to 3000 pages to go with it. Just don’t go beyond 24,000 words if you want Gemini to remember the whole chat before it responds. 

Google Calendar Trick

When I asked a podiatrist’s office to send me my X-rays, I got an authorization form that asked me the date of the diagnosis. Gosh, who remembers that kind of thing? So with my Google calendar open on my screen in the Google Chrome web browser, I clicked the “Ask Gemini” button. It instantly found my October 2025 date. This trick might be useful for finding anything you’ve added to your Google calendar, such as the last time you had your neighbors over for dinner.

Ad Blockers

In the past, when a website asked me to disable my ad blocker, I couldn’t find one. But Google Chrome has one built in. To unblock the ad blocker in Chrome, type the following at the top of your screen, without the quotes: “chrome://settings/content/ads”. When it comes up, check the first bulleted item, which reads “Any site you visit can show any ad to you.” 

At Forbes.com, I was able to read some of their articles for free because I was no longer blocking their ads. That’s how I learned that the US has a record 144 immigrant billionaires worth a collective $2.2 trillion.

Signing Up for Stuff

If you’re creating a new account at a sketchy site, a spammer can copy both your username and password. But they will only get your email address if you sign up with your Google or Facebook credentials. To avoid either problem, it’s best to create junk email addresses. I created some at AOL, Yahoo and elsewhere. But I wish I’d remembered to use them when registering on trivia sites. Now I’m bombarded with trivial email.

What Movies and Series to Watch

“Letterboxd” is both an app and a website. To start, open the app or site and click on the posters of movies you like from a list of what they call the top 500. This leads to more tailored lists. That’s how I found “Comfort Movies.” Clicking on one, I looked to the left to see whether it was available on Netflix, Amazon Prime or elsewhere. 

Man in airplane cargo

Podcast on Crime

My nephew’s wife recommended the “Criminal” podcast from ThisIsCriminal.com. I just listened to episode #375, “Mr. Apology.” 

In 1980, posters appeared in subway stations and on telephone poles in New York City with a phone number to call. When you called it, you would hear a message from an answering machine, a novelty at the time. The voice would announce: “This is Apology. Apology is not associated with the police or any other organization but rather is a way for you to tell people what you have done wrong and how you feel about it.” One guy admitted to peeing in a trashcan inside a bank, he was so desperate. Another admitted feeling badly that he’d embezzled $400,000 from a bank, but said he planned to do it again because he liked living like a king. A mailman apologized for tossing out the mail he was too tired to deliver. A little girl apologized for throwing sticks at a dog that was menacing her. A guy admitted to stealing toilet paper at work. I’m looking forward to the episode on a crime involving a volcano.

Online Notary

Everything, it seems, is now online. Recently, for example, when using an online notary service called “Proof,” a notary popped up on my screen in literally one second and a second witness arrived the instant she was called. Of course, I had to answer some security questions beforehand, and upload my ID, but after that, my document was instantly notarized. 

Proof, OneNotary and Notary Live charge $25 per seal for pay-as-you-go notarization, and OnlineNotary.net charges $20. All four services operate 24/7. 

Targeting Cancer

What if you’d been throwing darts at a tiny dartboard, unaware that the real dartboard was next door, and enormous? That’s what happened in cancer research. A quarter of all cancer drugs in development go after just 38 targets, according to a 2025 industry analysis. But scientists at RyboDyn, a San Diego-based biotech, just found thousands more to aim at, and discovered a new drug that killed cancer cells in a lab dish. They seem to have invented a whole new biology. 

Internuts

–Darioamodei.com has essays by the CEO of Anthropic. He says that we are closer to real danger with AI than we were a few years ago, but  we should avoid believing that doom is inevitable. It’s both false and self-fulfilling.

–“Solar Trees Could Save Forests From Deforestation While Generating the Same Power as Solar Farms.” Search on that phrase to discover that solar trees preserve 99 percent of the forest, compared to just 2% otherwise with traditional methods.

Filed under: AI, business, Google, Internuts, inventions, science, spam, video, web | Comments Off on FROZEN GEMINI

Recently, I went to a big chicken dinner at my sister’s house, where I happened to be staying. The issue was: “What shall Health Nut Joy have?” So I asked Microsoft Copilot.

First, I told my artificially-intelligent sous chef about the stuff in the pantry and fridge. Then I asked it to come up with an oil-free, salt-free, sugar-free, vegan recipe. It gave me three choices.

The first was “Italian Broccoli–Brussels Sprout Pesto Bowl.” The second was “Mexican Sweet Potato–Black Bean Tacos.” The third was “Moroccan‑Style Sweet Potato & Black Bean Tagine.”

There was one complication. I had to make my dish early, or my brother-in-law, a true chef, wouldn’t have the time or space to create his own masterpieces. Copilot said that the first two recipes could turn soggy, limp, or sulfur‑y by the time we sat down. But the Moroccan recipe would be great. It was right. The bride’s Bangladeshi mother was delighted to have something spicy and I made it for myself four more times that week.

To create your own AI masterpiece, go to Copilot.com, or another form of AI. ChatGPT, Claude and Google Gemini are great too. At gemini.google.com, I typed: “Make me a different international dinner every day with no repeats.” Its chitchat back and forth taught me that lime zest and fresh basil can substitute for Makrut leaves. I also learned that hemp seeds are the best substitute for coconut. 

Removing a Crazy Lady

Eighteen of us were lined up at my nephew’s wedding reception, wearing our best smiles. But just before the photographer snapped a photo, a chic lady jumped in beside us. “Was she the wedding planner?” I wondered. After invading the photo, this youngish beauty disappeared into her chauffeured Rolls Royce, laughing her head off.

I thought of using Google’s Magic Eraser to get rid of her. It’s part of the Google Photos app. But my brother-in–law showed me how to do it in Apple Photos. As with Google Photos, it was a snap to make the crazy lady vanish.

Here’s how: On an iPhone, open the image. Next, tap the menu icon, three stacked lines. Select “clean up,” which is just to the right of “crop.”  Now, using your finger or a stylus, tap, brush or circle the image you want to delete. Wait a couple of beats. The computer will highlight the image it thinks you want to remove. If that’s it, select it and watch it go poof. If it’s wrong, just tap “reset” and try again. When the object is gone, tap “done.” 

On an Android phone, open the image, then tap the two stacked lines. Next, tap “Actions,” and choose “Magic Eraser.” Then use your finger or a stylus to tap, circle or brush what you want to remove. 

It’s not perfect. Both Android and iPhone did a sloppy job removing a guitar case at the feet of a woman playing in Central Park. But most of the time, it works great.

Free Game

As if we don’t have enough word games yet, here comes a new one, free from Merriam Webster. Start with seven letters, some of which are already in the right spots. Your mission is to add the other letters in the right order to form a phrase corresponding to a clue. If you do, you go on to the next level, with a different solution but the same seven letters. When I tried it, the first solution was “trip on a root.” The clue was about something that could happen in the woods. Look up “Merriam Webster Tagline” to find it. 

AI Alone in the Lab

What if artificial intelligence could step into a virtual lab and do everything by itself? That’s the promise of “Discovery Loop,” a company founded by Jeff Dean, one of Google’s early masterminds, who recently left Google. He’s trying to automate the whole process of scientific discovery. Just like driverless cars, we could have driverless experiments. 

At Discovery Loop, supercomputers will run thousands of virtual experiments simultaneously. But can they mimic the real world of science? In the physical world, biology refuses to be rushed. Cells take days to grow, clinical trials take years, and wet lab samples can get clogged. While AI can predict potential drug candidates overnight, living processes can’t be compressed into microseconds.

But the sheer speed of virtual experiments makes them promising. A discovery that could have taken decades could happen overnight. That’s why Alphabet and major Silicon Valley venture capitalists have designated the company a “Public Benefit Corporation.” That means that if the stock price drops because the company invested heavily in long-term research instead of near-term gain, shareholders cannot sue the leadership for failing to maximize profits.

Internut

“Dr. Dean Ornish: 71% of Alzheimer’s Patients Improved Without Drugs.”  Search on that phrase to  learn about an experiment with early stage dementia and Alzheimer’s patients. These were people who can’t watch a movie because they can’t follow the plot, can’t read a book because they’ll just re-read the same paragraph, and so on. The group was split into an intervention group and a control group. The control group got worse. The intervention group got better through lifestyle changes, including plant-based eating, exercise and social connections. This was proven by blood tests that look for specific Alzheimer’s biomarkers.

Filed under: AI, games, Internuts, photos, photoshop | Comments Off on COOKING WITH AI

So much for free Wi-Fi on a plane. After an Alaskan Airlines flight, my laptop was overwhelmed by spams and scams.

As the pop-ups filled my screen, I saw a rotating parade of horribles, with messages like “TROJAN found,” and “Remove the Virus Now! I wasn’t sure what to do, except to avoid clicking on anything.

So I turned to Anthropic’s artificially-intelligent Claude AI for help. It suggested using Windows Defender for a virus check, then doing a malware check using the free version of Malwarebytes. I was surprised to learn that the obnoxious banners were the only problem.

After trying a couple of Claude’s suggestions, I chose the simplest one: blocking all notifications from all websites. To accomplish this, I went into the Notifications Settings page in Chrome and Edge and turned off “Ask before Sending.” Though it sounds counterintuitive, it stopped the spam.

But what about the next time I want to use free WiFi on a plane instead of the airline’s own entertainment portal? I’ll need a Virtual Private Network (VPN), I decided.

A free trial of the paid version of ProtonVPN seemed like the best choice.  Since it’s subject to strict Swiss privacy laws, it earns the highest trust of any VPN. It’s also considered the best free option.

But Proton VPN can be slow. In fact, I gave up before a British website loaded. Compared to other VPNs, Proton uses extra verification steps and privacy checks. Plus, its network is smaller, so a connection takes longer when a signal has to travel between distant nodes.

NordVPN makes more sense for frequent travelers like me. Its global network is huge and it’s good at picking the closest, fastest connection. Unlike Proton VPN, Nord doesn’t have a free option, but I wouldn’t want that anyway. Free VPNs don’t protect you against malicious attacks. They only make you invisible on the web. 

So I signed up for NordVPN Complete. When I got it, it was selling for $5 a month on the two-year plan. It not only blocks malicious websites, it stops drive-by downloads, filters out phishing pages that mimic login screens, and much more. It comes with a password manager, a terabyte of encrypted cloud storage, dark web monitoring and scam call protection for your phone. You can connect ten devices at once to NordVPN, and an unlimited number of devices can use the included NordPass and Nordlocker apps. I later noticed that NordVPN made it much slower to save a Word document as a PDF, but I can always turn it off when I’m using my home WiFi.

Swimming with your Watch

On vacation in Southern California, I swam every day with my Apple Watch. But long swims in the salty bay or ocean aren’t good for it, I learned later.

Fortunately, however, Apple Watches have much better protection against liquid damage than other smart watches. They can eject water to keep things humming even after you goof up. After four days in a row of bay swims without rinsing, I tapped the exercise icon on my watch, then tapped again to start a fake swim. This automatically turned on the water lock feature, so I was able to rinse my watch for the recommended 30 seconds, which helps to prevent corrosion. Then I pressed and held the crown to eject the water.

Blocked Contacts

 “If you block someone you previously trusted, will they know it?” I wondered this because I had recently expelled a friend from email, text and phone. I did it in the “Contacts” app.

If the former friend tries calling my iPhone, he will hear my phone ring once, twice or not at all before getting an option to leave a voicemail. If he leaves a message, it will appear in the Blocked Messages folder at the end of my Voicemail list. So far there’s nothing there, which means he didn’t leave one.

On Android phones, which vary according to the manufacturer and carrier, the call will either be rejected or sent to voicemail, where it might appear in a “blocked calls” folder. Alternatively, instead of blocking a person in your Contacts app, you could use your carrier’s portal, which will send a message to the person that you are not accepting calls.

Text messages work differently. On an iPhone, the blocked sender won’t see a “delivered” or “read” message after they send their text to you. Nor will you ever see their text. But on Android, you can check the “Spam and Blocked” folder.

On Facetime, the person will hear a phone ringing indefinitely without ever connecting. You won’t know they called, and they won’t get a message saying you’re not available. 

Internut

“Meet the robot trained on millions of internet videos.” Search on that phrase to watch an amazing video clip.

For years, robots in factories have been doing assembly line work, but when things went awry, the robots got confused. So now a company called Rhoda AI has trained them to expect the unexpected and react appropriately. A ZDNET video shows a robot doing intricate tasks for a German automaker, after being trained for only about 15 hours. That’s because the training data is continuously refined and sifted from hundreds of millions of public videos, until only the key moments have been saved to provide training for a precise task.

Filed under: AI, gadgets, security, wearables | Comments Off on THE DANGERS OF FREE WIFI

I just took my first Waymo, Google’s self-driving taxi. When a jaguar pulled up, I knew it was the Waymo because of the cameras and radar on top.

I struggled for a second to unlock the door. Then I realized I had to tap “unlock” in the Waymo app on my phone. A cheery female voice greeted me as I sat down, encouraging me to press the support button whenever I wished. Instead, I asked Google Gemini on my phone. It told me how to play my favorite podcast on YouTube Music instead of the New Age-y stuff coming from iHeart Radio. 

My Waymo proceeded cautiously, going slower than I would have as we rounded the first turn. Possibly it saw something I didn’t. Statistics show that self-driving cars are safer than human drivers. If I’d been in doubt, I could have tap “pull over.”

It took longer to get to San Francisco International Airport than expected, because Waymos don’t have freeway access everywhere yet. But I enjoyed checking out the features. I could even have added stops, but if I didn’t get back in the car within 30 minutes, it would have gone on without me.

As we approached the airport, the invisible driver told me it was looking for a place to pull over. It reminded me to check for belongings and to pull the handle twice to exit. I couldn’t help waving goodbye. The cost: $62.40, no tip.

What about privacy? The interior cameras are always on. They help the AI bot answer your questions. They also check to see that the car is clean, confirm that you’re wearing your seatbelt and register lost items. Under certain circumstances, a support team might review the video even as it’s being recorded. Recently, for example, in San Mateo, California, two 15-year-olds were allegedly drinking and shooting toy guns from a Waymo. The company’s systems flagged the behavior, disabled the car, and called police.

Anything audible is not recorded, however, unless you press the support button to ask a question or deliberately turn on the microphones. So “sing out,” as my Waymo suggested. If “Kumbyya” comes on the radio, I might just do that. For more information, Google the online tutorials.

Creepy Glasses

I wouldn’t notice if someone’s smart glasses were recording me because I’ve never been aware of the red indicator light on the glasses. Heck, I don’t even know anyone nearby who has a pair. But Gen Z is hopping mad and they’re not going to take it anymore.

Modifiers, nicknamed “modders,” are tweaking smart glasses to prevent the recording light from coming on. This means you could be videoed or photographed in a public bathroom or elsewhere without your knowledge.

In protest, young people have put up large posters at bus stops in London, including one just a few blocks from a Meta Ray-Ban shop. The ad shows a regular pair of Meta Ray-Bans morphing into a zombie with the tagline “We’re always watching.” Perhaps in answer to this, Meta is rolling out a “v26” firmware update that will block the glasses’ camera if the indicator light has been tampered with. 

But Meta is also reportedly working on a prototype with a “super sensing” mode. “Super sensing” allows the glasses to continuously record audio and snap photos every few seconds. It’s supposed to make it easier for users to ask Meta AI about anything in their environment.

There are some serious privacy concerns, however. Though Meta says that the footage is only processed by AI, the company has previously allowed contract workers to review it, even if the user has opted out of data sharing. So their slogan, “processed only by AI,” may be more of a sales gimmick than real protection. 

Here’s the second problem. Bystanders don’t give permission to be continuously recorded, and  probably won’t know when it’s happening. Thirdly, you can’t access the recordings and photos taken by your own glasses, so it’s impossible to inspect, export, correct or delete them. Unknown footage might draw you into legal battles over images you didn’t know existed. Finally,  even if Meta deletes the raw data, they could first extract precious info from the stream, including locations visited, faces encountered, conversations overheard and routines executed. These could be retained and analyzed as metadata long after the raw video has been discarded.

Watching the Planes Come In

Go to FlightRadar24.com to watch the planes come in. Or download the free app.

Once you’re in, just enter an airport’s call letters into the search box. I typed “ORD” for Chicago O’Hare International, clicked “show on map,” and saw a whole fleet of planes flying over Lake Michigan, each represented by an image. Clicking on one of the animations, I learned that it was a plane coming from Madrid and arriving almost an hour early. The information panel on the left showed the flight number, airline, altitude, speed, aircraft type and more. If you click on the 3D button in the lower left, you’ll see a 3D view of the flights. 

The free version gives you only three 3D sessions, but you can get unlimited 3D and other features with a premium subscription. I watched a plane arriving in Chicago from Orange County. But it’s just a flight simulator. You don’t see live video of the actual planes.

Filed under: AI, apps, gadgets, web | Comments Off on MY FIRST WAYMO