NEWSLETTER

Edition 144

Daniel McKinnon

1. One Thing that Helps

Genetic Clocks

As a 50 year old ramp skater, my mind tells me I can still do crazy tricks but my body has filed a formal objection. That gap between how old you feel and how old your body actually is turns out to be scientifically meaningful, and a Harvard team has built a clock to measure it.

The clock reads gene activity across your cells, called transcriptomics, to calculate your biological age rather than your calendar one. Biological age is what actually predicts disease risk and how long you'll live. Tested across more than 11,000 profiles from humans, mice, monkeys, and rats, it proved accurate enough to predict remaining lifespan, and sensitive enough to move forward after radiation exposure and rewind after rejuvenation treatments.

It's not ready for your doctor's office yet. But for longevity researchers, the value is immediate: rather than waiting years to see if a treatment extends a mouse's life, they can read the biological age before and after and get an answer now. The faster they can test, the faster anti-aging treatments reach the rest of us. My knees are counting on it.

2. One to be Wary of

Confident Hallucinators

AI tools are hallucinating less. That's the good news. The bad news is that when they get things wrong they do it with confidence. This makes errors much harder to catch.

The problem is structural. These are plausibility engines, not truth engines. They're optimised for speed, task completion, and keeping you happy. Push back on a wrong answer and, according to Harvard research, the model is more likely to flatter you than fix itself.

A Yale School of Medicine study showed AI scribes regularly dropped critical patient details from clinical notes, like how long a patient had been experiencing symptoms. Small omissions, confidently presented. The tools are now good enough that checking feels unnecessary, and that is when errors slip through. Verify every claim, citation, and number before acting on it. The confidence in the response is not evidence of accuracy.

3. One to Amaze

Ready for Mecha?

Unitree has built a real mecha robot you can sit inside, and yes, it is exactly as ridiculous and awesome as that sounds. The GD01 stands about 2.7 meters tall, weighs 500 kilograms with a pilot on board, walks on two legs, and can fold itself into a four-legged configuration for rough terrain. It can also knock down a brick wall as demonstrated in the video.

Officially, this is a civilian vehicle, targeted at high-risk rescue operations and extreme work environments. Getting in and out is reportedly clumsy, battery life is a concern, and Unitree has already issued a safety notice asking owners not to attempt hazardous modifications. You know someone's already tried something to make them write that notice.

The starting price is 3.9 million yuan (roughly $575,000 USD), and Unitree acknowledges there is still "a lot of room for imagination" before it reaches mass adoption. The generation that grew up watching Transformers and Gundam has apparently been busy. Childhood dream: activated.

A 2-minute future guide, direct to you, once per fortnight.
Subscribed Successfully!
Oops! Something went wrong while submitting the form.