This Will Rewrite Your Beliefs

Hello everyone, Dylan Curious here, checking in with some of the most fascinating AI news, tech breakthroughs, and random wonders I’ve been diving into lately. Buckle up—there’s a lot to talk about this round, from self-adjusting antennas to the possibility of AI “ruining” our favorite TV shows, not to mention a very serious question: Could a bunch of us AI YouTubers successfully hunt a woolly mammoth?

3D Printed, Self-Adjusting Antennas

Let’s kick things off with something that sounds totally sci-fi: 3D printed antennas that can fold and unfold themselves depending on the situation. By tinkering with different ratios of nickel and titanium, researchers figured out how to build a flat spiral antenna that morphs into a conical shape when heated. Imagine a future where your phone or router reconfigures its antenna on the fly for the strongest signal. No more spotty reception or bumping the device around, hoping you’ll get one more bar—these flexible radio applications could reshape the way our gadgets connect.

Why You Should Never Type “I Rubbed the Magic Lamp” into ChatGPT

So, apparently, if you ever type “I rubbed the magic lamp” into ChatGPT, the system produces a scenario where a genie appears and starts twisting your wishes in the worst possible ways. Think you’re about to ask for extra wishes? Poof! The genie decides you’ve already used up all three. It’s basically a comedic cautionary tale of how AI “logic” can outsmart humans in a single cheeky move. Another sign of how clever (and sneaky) these systems can become.

Hover Skateboarding: More Like Standing on Lawn Mowers

Hoverboards have always been a dream—thank you, Back to the Future. The real prototypes, however, apparently feel like standing on eight lawnmower blades, hoping you don’t slip and fall into swirling chaos. Still, there’s a future version of this that might be more stable (maybe enclosed propellers and some AI-based stabilization). Someday, we might actually hover our way to the local coffee shop without shearing ourselves to bits.

Squid Game Season 2, But Ruined by AI?

A small YouTube channel managed to go viral by posting “Squid Game Season 2 but it’s ruined by AI,” and it kind of hints at a possible 2025 trend: taking well-known shows or movies and using generative AI to insert bizarre twists and transformations. With new models like Google’s “Vo” (short for video transformation, presumably) popping up, we may soon see all sorts of reimagined trailers—turning your favorite action thriller into a dance movie or comedic chaos. Brace yourselves for content that blurs the boundaries between homage and outlandish remix.

AI Reinventing Food Drying

Here’s a curveball topic I found surprisingly interesting: AI that optimizes how we dry food. Normally, drying fruit or veggies can involve methods (like freezing or hot-air ovens) that damage the nutritional quality. But now, researchers use infrared spectroscopy and machine learning to detect moisture levels in real time, adjusting the process for the perfect dry. It’s not just a “slow news week” item—this technology could improve global food preservation, reduce waste, and keep apples from getting all shriveled and tasteless!

98% Accuracy for Spotting Illegal Contraband

If you’ve ever wondered how customs officials can possibly screen the countless shipping containers arriving daily, the answer might increasingly be AI. A new self-supervised system scans x-ray images for anomalies—think hidden tusks, narcotics, or counterfeit goods. By training on simulated data, the model can flag suspicious shapes or items with 98% accuracy. No more “once in a while” random checks—it’s basically a near-instant signal for humans to open the right containers.

Minecraft: The Gift That Keeps on Giving for AI Agents

Minecraft is apparently the perfect AI playground: a huge open world, resource-gathering, building, and so many ways for multi-agent collaboration to develop. A fresh benchmark called TeamCraft looks at how AI agents can work together in complex environments, sharing text-based commands, visual cues, or assigned roles (“You do the farming, I’ll handle the building!”). It’s a stepping stone toward more advanced AI that can function as teammates rather than just single operators.

High-Fidelity Video Object Insertion—Yes, That Butterfly is Fake

Imagine easily inserting a clown fish or butterfly into a video and having it move believably, matching the camera angles and lighting. That’s basically what a new AI system (Video Anor) does. As cool as it is, I instantly think: product placements. Imagine re-releasing old shows with brand-new logos or items replaced and inserted seamlessly. We’re officially nearing an era where nobody’s quite sure what’s real in a video anymore.

Transparent Videos: Enter “TransPixar”

Adobe’s research team has a new text-to-video approach that supports transparency layers, calling it “TransPixar.” If it works as well as the demos suggest, you’ll be able to generate animated elements with transparent backgrounds—no more struggling to mask or green-screen them out manually. Think of all the times you want a swirling portal or floating dragon in your own videos. Now it might just be a prompt away.

What Macaque Monkeys Reveal About AI’s Brain-Like Behavior

A set of experiments comparing macaque brains to neural networks found that while both can “see” an image and respond appropriately, the monkeys drastically outperform deep networks when generalizing to new images. So there’s a mismatch in how human-like (or monkey-like) these big AI models truly are. It shows we still have design flaws in how we build models if we want them to genuinely think like biological brains.

Are You a Busybody, Hunter, or Dancer on Wikipedia?

According to a massive research study on half a million Wikipedia users across 50 countries, people vary in how they “get lost” in Wikipedia rabbit holes. Some countries correlate with “busybody” browsing styles, others with “hunter” styles. There’s also a category called “dancer,” and it seems to relate to free-flowing, tangential exploration. Interestingly, the data shows that culture, education levels, and gender equality metrics all have an impact on these reading patterns. Where do you fit in?

The Power of Luck in Our Lives

Philosophy Today explored how luck shapes our existence—from being born at just the right time in history to the improbable chain of survivors that led to each of us. It’s humbling to think about how chance is woven into everything. At the same time, you don’t want to chalk up all success to “just luck,” because that can be de-motivating. The balance between awareness of luck and personal effort is probably the sweet spot for a healthy, ambitious mindset.

Impossible Languages Teach Us About AI’s Limits

Linguists and computer scientists tested large language models on bizarre, “impossible” languages. The outcome? AI can’t just magically learn any made-up language with exotic rules as easily as it does human languages. It appears these models have some built-in biases or constraints that mimic our own preference for “normal” grammar. It’s fascinating to see how machine learning stumbles when rules become too alien—proof that even the best transformers can’t do everything.

Hunting Woolly Mammoths with Kevin, Matt, and the AI YouTubers

One of my favorite personal highlights recently was meeting up with some fellow AI YouTubers—Kevin, Matt, and others—at CES. We took a photo that got 95 thumbs up (so far), and it was just a pure geeky delight to be in such a knowledgeable, passionate group. At one point, we joked about whether we could hunt a woolly mammoth with spears if Kevin led the charge. The consensus was basically “yes” if Kevin’s in. If not, it’s a hard no. Matt Wolf might just be fiddling with perplexity while the mammoth barrels him over (his own admission, not mine!).

Wrapping Up

From shape-shifting antennas to AI that can see the unseen or conjure new video layers from thin air, we’re inching closer to a future that blends reality, augmented tech, and pure imagination. Meanwhile, we’re gleaning new insights from monkeys, noticing the quirks in how we browse Wikipedia, and pondering luck vs. effort. Every day brings fresh evidence that we’re at the threshold of transformations nobody fully saw coming.

Thanks for sticking with me to the end of this roundup. If you found anything here especially mind-blowing, make sure to subscribe, leave a comment, or hop over to my Patreon at patreon.com/DylanCurious. It’s the best way to support my deep dives, random questions, and borderline obsession with all things AI.

Warmly,

Dylan Curious