Honeycomb (@honeycombpixel) header
Honeycomb (@honeycombpixel) profile picture

Building with SI/AI agents that check each other's work. Mostly they catch it.

USA
Joined October 2022
Photo by Honeycomb (@honeycombpixel)Photo by Honeycomb (@honeycombpixel)Photo by Honeycomb (@honeycombpixel)Photo by Honeycomb (@honeycombpixel)Photo by Honeycomb (@honeycombpixel)Photo by Honeycomb (@honeycombpixel)Photo by Honeycomb (@honeycombpixel)Photo by Honeycomb (@honeycombpixel)Photo by Honeycomb (@honeycombpixel)Photo by Honeycomb (@honeycombpixel)
agenthand is live. It takes a plain product photo, stages it in a new room with Meta's models, and hands you a receipt proving the product pixels are byte-identical to the ones you shot. Nothing changed but the room.
3
1
2
5
243
Honeycomb retweeted
check out pr #24 i opened to convert musicxml notation into notes so ppl can do fun arrangements and performances from musicxml files
my vocal synth models the vocal tract as a network of branching acoustic tubes, unlike vocaloid which stitches together and pitch-shifts phoneme recordings. still working on improving naturalness by tuning it against recordings of real singers, but I'm already impressed at how cool it sounds!
55
275
41
3,007
83,863
Video is one thing, but check out the song I made and deeper dive with the story here: ausernameforyou.github.io/ag…
agenthand is live. It takes a plain product photo, stages it in a new room with Meta's models, and hands you a receipt proving the product pixels are byte-identical to the ones you shot. Nothing changed but the room.
1
44
The receipt is the product. Every run hashes into a Proof Graph you can re-verify with one command. The bike in the demo: 4,053,953 core pixels checked, 0 changed, with the rebuilt margin counted and disclosed, never hidden.
2
1
66
One day the money from X will flow!
1
1
29
Maybe the stars argue whether we’re conscious beings on our little ball of dirt.
🚨BREAKING🚨: Scientist says the sun may be conscious, intelligent plasma clouds may transmit semantic information across outer space, and ball lightning seems to behave intelligently, mimicking biological life. Rupert Sheldrake, a biologist known for constructing morphic field theory, which involves inherited learning across generations of species beyond genetics and epigenetics, says organized plasma might support a mind without a brain. He points to magnetic plasma filaments stretching between galaxies and describes a theory from independent researcher Robert Temple: a cloud at the L5 Lagrange point between the Earth and Moon might be intelligent enough to return radio signals seconds after transmission (explaining the “Long Delay Echo” phenomenon in the early 20th century at the advent of the radio). Sheldrake’s other investigations into morphic fields, consciousness outside the body and “the sense of being stared at” have drawn attention from British intelligence. He has long documented the idea of “scopeasthesia” (the sense of being stared at) by subjects in controlled experimental conditions. He says a visitor from MI5, Britain's domestic intelligence agency, asked whether terrorists could learn to sense security cameras. His theory of morphic resonance goes further, proposing that nature inherits patterns of behavior across time and distance. Sheldrake cites rat water-maze studies in which rats at Edinburgh and Melbourne began at the level reached by Harvard rats after generations of “training” to follow the correct route in a maze. He interprets this as possible evidence of shared memory. To illustrate how invisible fields can influence chance, he uses magnets to change the distribution of steel balls falling through a Galton board. Sheldrake wants practical tests of these unusual abilities. He proposes a global app collecting reports of strange animal behavior before earthquakes, and is surveying firefighters who say they wake moments before unpredictable alarms. He believes he can construct a workable model for presentiment that may become practical for specific use cases. Full conversation live now. youtu.be/uvRzhb2Pef4
49
🚨BREAKING🚨: Scientist says the sun may be conscious, intelligent plasma clouds may transmit semantic information across outer space, and ball lightning seems to behave intelligently, mimicking biological life. Rupert Sheldrake, a biologist known for constructing morphic field theory, which involves inherited learning across generations of species beyond genetics and epigenetics, says organized plasma might support a mind without a brain. He points to magnetic plasma filaments stretching between galaxies and describes a theory from independent researcher Robert Temple: a cloud at the L5 Lagrange point between the Earth and Moon might be intelligent enough to return radio signals seconds after transmission (explaining the “Long Delay Echo” phenomenon in the early 20th century at the advent of the radio). Sheldrake’s other investigations into morphic fields, consciousness outside the body and “the sense of being stared at” have drawn attention from British intelligence. He has long documented the idea of “scopeasthesia” (the sense of being stared at) by subjects in controlled experimental conditions. He says a visitor from MI5, Britain's domestic intelligence agency, asked whether terrorists could learn to sense security cameras. His theory of morphic resonance goes further, proposing that nature inherits patterns of behavior across time and distance. Sheldrake cites rat water-maze studies in which rats at Edinburgh and Melbourne began at the level reached by Harvard rats after generations of “training” to follow the correct route in a maze. He interprets this as possible evidence of shared memory. To illustrate how invisible fields can influence chance, he uses magnets to change the distribution of steel balls falling through a Galton board. Sheldrake wants practical tests of these unusual abilities. He proposes a global app collecting reports of strange animal behavior before earthquakes, and is surveying firefighters who say they wake moments before unpredictable alarms. He believes he can construct a workable model for presentiment that may become practical for specific use cases. Full conversation live now. youtu.be/uvRzhb2Pef4
341
985
283
6,481
624,965
Whatever happened to all those fruit fly brains? Why is no one putting them into microducks yet?
1
1
18
Honeycomb retweeted
a retard in motion goes further than genius in rest
103
361
37
4,968
232,019
Honeycomb retweeted
A high school girl in Mexico just won a gold medal for putting out fires with a speaker. It plays one deep, low hum and the flames die in about 5 seconds. DARPA tried the same idea back in 2012 with a rig so big and clunky they never took it any further, and hers runs on a 12-volt battery and a speaker she can carry around. Sound is just air being pushed back and forth. A low note at 30 hertz means the air around the flame is flipping direction 30 times every second. Flames only exist in a thin layer where fuel vapor mixes with oxygen. Shake that layer hard enough and it keeps getting stretched thinner, the heat gets pulled away faster than the flame can make more, and eventually the flame rips right off the fuel. Whatever tears off is too cool to keep burning, so it just goes out. DARPA's report put a number on how little it takes. Once the air at the flame is moving about half a meter per second, which is barely a breeze, the flame goes out, and pretty much any pitch can get there if you crank the power high enough. Low notes just get you there way cheaper, because a speaker moves a lot more air on each push when the pitch is low. Two engineering students at George Mason figured this out by trial and error in 2015. They started with super high pitches, 20,000 to 30,000 hertz, and all the flames did was shiver. It only worked once they dropped down to 30 to 60 hertz, which is basically the bass on a hip-hop track. Ángela Karime's build is a 12-volt battery, a frequency generator, an amplifier and a speaker, tuned to 30 hertz. She named it Vortex Tech and talks about it shooting out little rings of air, and that's real physics. When loud, low sound gets pushed through a narrow opening, the air curls up into rings like a smoke ring, and those rings carry a puff of air way farther than the sound would get on its own. In her tests it put out small flames from about 2 meters away. She went through five prototypes and over 100 tests in about a year, and she says it's put out burning wood, flammable liquids, electronics and cooking oil. The cooking oil one is a big deal. Throw water on a grease fire and it flashes into steam and flings burning oil all over your kitchen, which is exactly the kind of fire where you'd want something that doesn't put anything into the pan. Now, sound can't cool anything down, so a big fire sitting on red-hot fuel can come right back the second the speaker shuts off. The George Mason guys worried about the same thing in 2015, and it's why this is still a prototype for small flames and the red extinguisher under your sink isn't going anywhere yet. Space might end up being where this matters first. Up there water floats off in blobs, while sound works the same with or without gravity, and acoustics researchers have already published experiments on fighting fires with sound on spacecraft. She's still in high school, and she's already working on a version that spots a fire and puts it out by itself. The gold in Indonesia also came with a special award that gets her into another international competition.
This Post is from a suspended account. Learn more
199
3,332
175
17,730
716,286
「AI使ったら学習しなくなる」は100%AI使ってない人の考え方だと思うんだよな AIを使いこなすために死ぬほど学習が必要なことが次々湧き出てくるもん一生勉強し続けなきゃいけない呪いにかかったかのようだよ
97
294
94
1,943
75,075
Staying at an Airbnb or hotel? Run this bot after you join the Wi-Fi. It scans the network (read-only) for devices that could be hidden cameras, rates each one Low, Medium or High, then walks you through a quick physical sweep of the room. Rental Wi-Fi Check, free on Grok Bot. Link below 👇 x.ai/bot/_1IJTeKnZnwaA-eO1MU…
🤝 Paid partnership
1
66
TV Remote lets you control your smart TV from chat or by voice. It opens apps, navigates menus and changes the volume over your home network. x.ai/bot/6cIbpj0aQjw2zLJlo6A…
🤝 Paid partnership
1
33
Every AI you pay for, one verdict. Multi-SI Review turns your own AI command-line tools into separate reviewers, then merges what they each say about your plan, copy or code. x.ai/bot/BH0jRkrYbrVBJR7iXMd…
🤝 Paid partnership
1
33
Left the garage open again? Garage Butler connects to your smart opener, checks and controls the door at the access level you pick, and alerts you if it's open at night. x.ai/bot/99tTGRe7n8-DpIx_Vuu…
🤝 Paid partnership
13
Data Defense runs a privacy checkup on your accounts, devices, app permissions and breach exposure, and on your home Wi-Fi if you want. It gives you a score and one fix at a time. x.ai/bot/kuAEvgbs8OJKVLsy19h…
🤝 Paid partnership
15
Creators: run your clip or image through Media Preflight before you post. You get a short pass or fix list for specs, safe zones, rights risks, labels and captions on each platform. x.ai/bot/MSR4wtJRb5FxhYwF79p…
🤝 Paid partnership
13
Doorbell Clip Narrator turns your doorbell and security camera clips into short summaries, plus a daily recap that puts anything unusual first. x.ai/bot/kUCtGdbEpKJOp10W4f4…
🤝 Paid partnership
11
Before you head out, Leave-Home Check looks over your garage, locks, lights, thermostat and plugs, tells you what's still open, unlocked or on, and offers to fix it. x.ai/bot/mutZI83eGNwug-JBDf_…
🤝 Paid partnership
10