Hacker Newsnew | past | comments | ask | show | jobs | submit | quietsegfault's commentslogin

The qwen flamingo looks like it’s smoking’ a doobie.

Which industry? Tech? Surveillance? Government? I know my father in law is a MAGA racist who believes whatever makes it easy to justify his own beliefs. I’m not sure you can ever reliably judge someone’s true motives in a professional setting.

My kids’ schools also are anti-AI. My kids don’t have personal experience with it yet, other than funny pictures on Gemini.

Can I get involved like an arms dealer and influence the course of chimpanzee history?

I look at it more like speedrunning a level. You're skipping the parts of the level that take up the most time, some times using hacks. Is it universally as much fun as playing the game? No, just like using AI to prototype might get you to the same place, but without the experience of discovery and blockers along the way.

I like to experiment with AI flows to make iteration quicker, then once something work investing in is found, back up and build something that's actually repeatable.

The photons arriving at the satellite from a ground target aren’t getting weaker due to distance in any way that defeats you, because the sun is illuminating the target and the NRO just needs to collect enough of the reflected photons.

SQUID sensors (the most sensitive magnetometers that exist) require magnetically shielded rooms to record cardiac signals at centimeter range.


They require a shielded room to increase the SNR. SQUID sensors are sensitive enough to record cardiac signals at distance. The issue is SNR.

What they are saying is that they produced a low noise sensor array and managed to increase the SNR through computation. They also stated that it was an ideal environment with no other electrical/magnetic interference.


> They require a shielded room to increase the SNR.

Not anymore. That is exactly the purpose of Darpa Ambient program: https://www.darpa.mil/research/programs/atomic-magnetometer-...

Demo from 2022: https://www.youtube.com/watch?v=VTnIXWCBYTw


AMBIIENT’s goal was biological imaging.. sensors near a body for medical/neuroscience applications. The range being discussed is still on the order of meters at best, not miles.

> AMBIIENT’s goal was biological imaging.. sensors near a body for medical/neuroscience applications.

As with all DARPA projects, there is a civilian use-case and a military use-case.

That demo was at a conference, in a city, surrounded by electronics and RF noise. The fact that it worked at all in that environment is surprising. As the subject got closer to the apparatus, the signal became larger than the background noise. So, I think the distance is limited mostly by background noise. The press release did state that it was an optimal environment for locating their target, i.e., an isolated person with only geomagnetic noise and known signatures of the aircraft.


If background noise was the factor, which it's not, but if it was, the background noise to combat would be the thermal noise floor of any physical conductor at any temperature above absolute zero. But that's not the factor. You're assuming the cardiac field is like a radio signal being transmitted rather than a local field effect. A magnetic dipole doesn't radiate energy outward the way a radio antenna does. It creates a static field that exists in the space around it, and that field geometrically collapses with distance. I ran the numbers on Wolfram Alpha, and at 1 meter from your chest the field is around 100 femtotesla. At 10 meters it's around 100 attotesla. At 5 kilometers it's around 10^-27 tesla.

Oh, but you say that you simply cool the sensor to 0K. Cooling helps, but you're still many orders of magnitude short even at near 0K, and you're doing this in Iranian mountains, not a dilution refrigerator.


You’re calculating based on classical mechanics. The sensors are optical-atomic magnetometers, using quantum mechanics to measure magnetic vector potentials (MVP), which behave differently and can produce seemingly non-local effects (Aharonov–Bohm effect).

> A magnetic dipole doesn't radiate energy outward the way a radio antenna does.

See Section 15-5 of https://www.feynmanlectures.caltech.edu/II_15.html#Ch15-S5

Veritasium has a video on the weirdness of MVP: https://m.youtube.com/watch?v=XKSjCOKDtpk

See https://pubs.aip.org/aip/adv/article/13/2/025127/2877320/Dif... for an experiment measuring MVP with an optical-atomic magnetometer.

> Oh, but you say that you simply cool the sensor to 0K.

These sensors do not require cryogenics and have been developed with sensitivities of 10s of fT/√Hz, so approaching the quantum noise limit.

Essentially, you can think of it as measuring the energy of a magnetic dipole in space instead of measuring a magnetic field.


Earth’s geomagnetic noise fluctuations are on the order of nanoteslas (10^-9 T), which is 18 orders of magnitude above the signal they claim to have pulled out.

It’s below the thermal noise floor of any physical measurement system that obeys thermodynamics. You can’t engineer around it because it’s not an instrumentation problem. The signal is smaller than quantum noise limits at that scale. “AI” filtering doesn’t help when there’s no signal to filter. You can’t computationally recover energy that isn’t there.

This is certainly bullshit of the finest, most grassy and odorous caliber.


What if they have a bunch of prerecorded data for the guys heart, then use a lock in amplifier to see if there is a signal in the noise

https://en.wikipedia.org/wiki/Lock-in_amplifier


It could be real. Or it could be a smokescreen for their remote viewing program. But isn’t the most likely explanation that pilots carry a radio/gps device and that’s how rescue found him?

Background noise is pretty random. A cardiac signal is pretty regular.

Could be like astrophotography. Where you focus long enough that the background noise cancels out.

Believe it or not? Also AI!

Earlier today I had Claude manage a TV show downloading system, and then it hacked a NES rom for me, setting up a web site so my kid could just go to a page to play it. So, I guess it’s not uniform.


Consider applying for YC's Summer 2026 batch! Applications are open till May 4

Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact

Search: