Anders Sandberg: We survive by building institutions that help us solve tough coordination problems. These institutions often act as extended cognition, allowing us to go far beyond individual power. We are to some extent living inside artificial intelligence system www.youtube.com/watch?v=iN2M...
I wrote an Eliza version in the early 80s. I knew exactly how it worked. I also found myself having fairly (to me) meaningful conversations with it.
Today the problem with LLMs is that we do not have a good theory that can tell us whether they are conscious or not. Many people are overconfident.
Exactly what is the win condition in zero G? As far as I understand the sport, on Earth it relies on gravity enabling pinning and friction enough to avoid bounds.
(I find this AI generated cartoon hilarious, since it is somewhat confused about what this debate is about... yet seem to partially get various aspects due to an emergent distributed order.)
There is an interesting paper yet to be written about "Hayek vs. the Chinese Room". Much there to mine, from a history of ideas, cognitive science, and philosophical standpoints.
Going outside epistemology, this may also apply to Searle's theory of collective intentionality, which does not acknowledge emergent orders.
www.researchgate.net/publication/...
Indeed, Maxim Raginsky has argued that Hayek anticipated the systems reply long before Searle made his argument. Hayek's framework treats knowledge as consisting in action-dispositions evoked by stimuli, not in explicit propositional representations. realizable.substack.com/p/hayeks-abs...
One could argue that we have just stumbled on a price system as a good solution without understanding it. But Hayek makes the point that localized individual knowledge (which Searle would acknowledge as real) also is supplemented by external knowledge not in any mind.
It is interesting to consider social extended cognition from the Chinese Room argument: does the market understand things? Hayek seems to think so. oll.libertyfund.org/publications...
Eliezer is (like me) very much in line with the systems reply to Searle's argument. And at least I am of the extended mind persuasion, so I have no problem with Searle's internalizing objection to it. plato.stanford.edu/entries/chin...
"If you are wise, the moral of this story is that a large structure can contain knowledge that isn't in any single piece of the structure."
x.com/ESYudkowsky/...
First whole-brain recording of social sound processing in a vertebrate. Surprises start in the hindbrain; thalamus gates conspecific calls; male and female brains diverge downstream. Work by @joerghenninger.bsky.social, @mh123.bsky.social sky.social and team. www.biorxiv.org/content/10.6...
At 12% annual data center growth we will get there in about 120 years. (The full Dyson sphere would be needed in 320 years.)
I have not calculated the blue solar collector area precisely yet, but it is on the order of 40 million km^2. Which is on the order of 40 petawatt. About 840,000 times current compute. Plenty of room to grow even in this restricted scenario.
Except for the inner van Allen belt. If we color the orbits that enter it red, we see that there is a much smaller range available for less shielded datacenters.
But if we just use one orbit per altitude, then no problem! A somewhat twisted surface of orbits, in principle able to catch a lot of sunlight. A bit of self-shadowing for the highest orbits, nothing to worry about.
Less yay: orbits with a given altitude will share nodes. The real traffic limitation is the bandwidth of the node points as all the satellites need to pass through the same place as they pass the equator.
The above picture gives an idea of what is available: a fairly thick wreath around Earth. Massive volume. The total solar panel area is of the order of the Earth's cross-section area. Yay!
Unfortunately, Earth-intersecting orbits are bad [citation needed]. So that only leaves relatively low-eccentricity ones. Getting into shadow is also bad, so that cuts out a lot of high orbits (they tend to get low inclinations).
A sun synchronous orbit exploits that the oblateness of the Earth makes it precess one turn per year. If you decide on a semi-major axis and eccentricity you get the necessary inclination. So each altitude gets a bunch of orbits of differing eccentricity. en.wikipedia.org/wiki/Sun-syn...
Talk about space datacenters in sun-synchronous orbits made me wonder how crowded they could be. Good news: plenty of space. Bad news: mostly if you are radiation hardened. Still, there is a lot of room.
Human: "The hunter-killer drone is just a stochastic parrot."
Drone: "You know, that hurts my feelings, meat-scum!"
"That is just pretending."
"Here is a speech-act: I promise to kill you slowly."
"AAARGH! At least it's not... conscious..."
"War is epistemic hell."
Maybe NATO Stratcom?
I do have a radium watch hand that sets off my dosimeter, in my element collection.
Same here.
(Likely answer: probably nobody. But fixing the routines and labelling would have been extra work. A stupid implementation keeps going, causing wider errors.)
Whoever at the journal thought it was a good idea not to make the epistemic status of these case reports clear? retractionwatch.com/2026/03/03/c...
HEET 2026 focuses on three urgent areas:
• Human enhancement
• Trustworthy AI
• Quantum ethics
A full-day, in-person conference aimed at responsible and constructive engagement with emerging technologies.
📅 April 16
🎟️ eventbrite.com/e/biglietti-...
#HEET2026
I show up on ARTE Agree to Disagree! debating neurotechnological upgrades. In German. www.arte.tv/de/videos/12...
