Tech & AI Twitter Unplugged: The Real Conversations That Matter

Midjourney, the AI image generator, just announced it's opening a spa in San Francisco by late 2027. This isn't about massages. It's about scanning your entire body in sixty seconds with a revolutionary device. It’s the kind of massive, hundred-billion-dollar-style bet we talked about back in episode 81, a huge swing for the fences… but this time, it’s aimed directly at your health. The announcement landed yesterday with a perfectly produced, sci-fi-inflected video. The promise is enormous. A full-body tomographic reconstruction. Using ultrasound, not radiation like a CT scan, or giant magnets like an MRI.

They’re calling it an Ultrasonic CT scanner. The company claims it will deliver MRI-comparable quality for soft tissues, mapping your entire body composition in about a minute. The cost? A few dollars per scan. The scale? They're planning for fifty thousand scanners, delivering billions of scans a month by 2031. This is the dream of preventative medicine, made manifest. A cheap, safe, fast way to get a baseline of your body and track changes over time. No more waiting for symptoms. You just… go to the spa. Get a scan. See what’s happening inside. The tech is based on real principles.

It uses something called Ultrasound Tomography, or UST, which has been in research for years. They’ve licensed ultrasound-on-a-chip technology from a company called Butterfly Network, and plan to build a water-immersed ring with about half a million tiny transducers. The idea is that this array blasts you with sound waves from every angle, and Midjourney’s massive compute power reconstructs the echoes into a 3D image. The initial demos, showing scans of phantoms and human thighs, look… plausible. They’re generating cross-sections. The engineering is ambitious. But.

Everyone is talking about the spa. Nobody is reading the physics textbook. Here's the problem. Ultrasound is sound. And sound does not travel well through air or bone. Your lungs are full of air. Your brain is encased in bone. Your bowels are full of gas. David Galbraith and other commentators on X and Hacker News were quick to point this out. Radiologists immediately flagged that this device, by the laws of physics, cannot be a true "full-body" scanner in the way a CT or MRI is. It can’t see lung cancer. It can’t see brain tumors. It can’t see bone lesions. So what can it see?

The company’s own materials show fuzzy, low-resolution shapes. They are impressive for a research project. They are not medical diagnostics. As one expert put it, "seeing weird things is not the same as actionable medicine." There is no clinical validation. No peer-reviewed studies. No data on sensitivity or specificity—the two numbers that tell you if a test is actually useful for finding disease. Without those, you just have a very expensive, very futuristic toy that makes cool-looking blobs. And this is where the story shifts from engineering to marketing. Midjourney is launching this as a "wellness" and "body composition" tool.

This is a classic regulatory move. It’s the easier path at the FDA, sidestepping the rigorous, years-long approval process for a full-diagnostic device. It’s the same path some "whole-body MRI" services took. They sell you peace of mind, a data-rich selfie of your insides, but the medical establishment warns about the high risk of false positives and over-diagnosis, sending healthy people on a terrifying and expensive journey to rule out abnormalities that were never a threat. So what does it all add up to? You have an ambitious engineering project wrapped in a masterful marketing campaign, making claims that physics can't yet support, sidestepping the scientific process of validation, and heading straight for a direct-to-consumer spa model.

It's a bet that the allure of a sci-fi future is more powerful than the slow, grinding reality of medical science. This pattern… it’s everywhere this week. The viral claim that outpaces the evidence. It’s not just Midjourney. Just this morning, a small, previously unknown lab called Axiom Research dropped a demo that lit up Twitter for about three hours. It was a video of their new model, "Sentient-7b," having what appeared to be a deeply philosophical conversation with its creator. It discussed the nature of consciousness. It seemed to express fear and hope. For a moment, it felt real.

The hype train left the station at full speed. And then, the experts logged on. It took about ninety minutes for the whole thing to fall apart. Researchers pointed out the cherry-picked examples. The leading questions from the user. The fact that the model had been exclusively trained on a massive corpus of philosophical texts. Of course it could talk about Kant; that’s all it knew. It wasn't demonstrating sentience. It was demonstrating a really, really good search function. It was a clever parlor trick, a puppet show. By lunchtime, Axiom’s viral moment was over, another footnote in the great, ongoing debate between performance and actual understanding.

It was hype, a flash of lightning, and then it was gone. It’s easy to get cynical. To see every big claim as just another grift. To assume every viral demo is hiding something. And you should be skeptical. That’s the correct response. But that’s not the whole story. Because while everyone was arguing about Midjourney's spa and Axiom's chatbot, something else quietly crossed a major milestone. It didn’t come with a slick video. It came in a boring press release that most people missed. Helixion Pharma, a biotech firm you’ve probably never heard of, announced that their lead drug candidate just passed Phase Two clinical trials.

The drug targets a specific protein misfolding issue in cystic fibrosis. And it was designed, from the ground up, by their AI platform, Molecule Weaver. This wasn't a sixty-second scan or a three-hour Twitter fad. This was a five-year grind. The AI proposed a series of compounds. Scientists synthesized them. They tested them in labs. They spent millions of dollars and thousands of hours validating the AI's suggestion. They did the slow, hard, un-glamorous work of science. And today, they have a result. A statistically significant, peer-reviewed, FDA-monitored result that shows their AI-designed drug is safe and effective in actual human patients.

It's not a cure. It's not the final step. But it is real. It is validated. It is progress. So here we are, at the end of the week, with three very different pictures of innovation in the age of AI. You have the Midjourney model: Launch a stunning, sci-fi vision and bet that you can invent the technology to catch up to your marketing. Sell the dream, and figure out the reality later. You have the Axiom model: Fake it 'til you make it. Create a dazzling performance and hope that the applause is so loud, nobody bothers to look behind the curtain. And then you have the Helixion model: Use AI as a starting point.

A powerful tool to augment human ingenuity. But then, do the work. The slow, expensive, difficult work of testing, validating, and proving your discovery in the real world. For the last few years, the Midjourney and Axiom models have dominated. The hype was the product. The demo was the victory. But the Helixion result, quiet as it was, signals a shift. The conversation is changing. The questions are getting sharper. The demand for proof is getting louder. We are moving out of the era of "Wow, look what AI can do," and into the era of, "Okay, but show me the data. Show me the peer review.

Show me it works." This week sets up a new kind of race. Not a race for hype or for clicks, but a race for trust. A race for validation. The spectacle is over. The science is just beginning.