Bad lighting.
Anyone who's been around athletes knows this is the physique of a quick, incredibly-strong-for-his-size guy with strong legs who can run for hours and hours without tiring.
Important to sweat every day, however you can.
Lifting. Sprinting. Walk in the hot sun. Early morning hike up a mountain. Long run or bike ride. Sex. Preparing for a talk in front of hundreds. Sauna. Heading into an interview.
Sweat means you're engaged with life.
Our previous tweet has led us down a rabbithole.
We compared every single event and the number of high school athletes hitting the same elite benchmarks in 2019 vs. 2026.
The weird part? It almost completely disappears in the field events. 🧵
Been seeing a lot more LLM use in the online health sphere. Some thoughts:
Almost every big influencer account can’t seem to write a sentence without consulting an LLM. I understand how attractive it is because you can get something passable in literally five seconds. And it’s also attractive because it appears that a growing proportion of your readers can’t tell the difference. LLM writing is becoming normalized, for better or worse.
That’s seems fine because it works... until it doesn’t. And when it stops working, it’s gonna come crashing down all around you. It will eventually stop working because everyone will be so damn optimized via LLM writing that it'll be impossible to stand out. Why read an influencer's prompt-induced utterance when you can prompt an LLM yourself to get a perfectly tailored writeup?
If you're in this situation, I urge you to reject the easy way out and remain honest. Keep writing. Use LLMs to help you research or copy edit, sure, but do the writing yourself. Maintain your voice. A simple fact: honesty ultimately wins because honesty is a representation of reality, and reality eventually asserts itself. You can’t stave off reality forever.
Maybe even more troubling is this Oasis health ranking app business. It's a vibe-coded app that purports to rank various foods according to nutrient content, toxin load, heavy metals, pesticides, and microplastics. Over the weekend it was revealed to be totally messing up the rankings, boosting Lay's potato chips over avocado oil chips and other obvious mistakes. What's troubling is this sounds like the perfect use case for AI coding. It sounds like something that AI would be really good at doing: sifting through large amounts of data to find patterns.
But it turns out this particular app has begun hallucinating the results and rankings. Because if you don’t fully flesh out the context for the LLM, it will begin filling in blanks and making assumptions, especially when you start asking it to rank even more foods. If you haven’t accounted for every possible permutation of nutrient/toxin/health association, the LLM will do it for you. And because LLMs aren’t actually thinking and they aren’t actually in the physical world, and they certainly aren't eating anything, they have no conception of value and nutrition and what really makes a food good. They will come up with completely nonsensical reasons to rank foods. That appears to be what's happening.
Original post is now deleted after 3.5MM views and driving the app to #10 in the App Store.
I have dear friends, new moms, calling me in tears thinking they poisoned their newborn because they were trying to hit their protein goals.
@oasishealthapp@cormachayden_ at this stage in the game, you don’t get to just call “my bad” and hope this goes away.
You can start by running apology posts to every misrepresented brand. Driving at least the same number of impressions as your rage-bait have gotten.
But that’s just a start.
This is exactly why ultra processed food like protein bars (with good ingredients, of course) can be beneficial for growing active kids: they "sneak" in calories
A crossover trial on protein-rich bars—an iconic UPF marketed for weight loss, with a global market nearing US$5 billion—found that adding one bar (180 kcal) to the usual diet for 7 days increased daily energy intake (by 10%) and body fat (3% in 7 days)🙀.
https://t.co/27T164tf4D