Fluffy Cyborg, IInd of the name
@fluffycyborgii.bsky.social
280 followers 340 following 2.3K posts
AKA Avel Guénin--Carlut Researcher in cognitive science, focusing on contextuality in life, mind and society; formalization & application of Active Inference; naturalizing ontology Occasionally schyzoposting on main. There are many occasions
Posts Media Videos Starter Packs
Pinned
fluffycyborgii.bsky.social
Hi 👋

Maybe it is time to give a proper impression of who I am and what I'm doing here.

First, I'm a PhD student / researcher in cognitive science, under the supervision of Andy Clark and Chris Buckley, financed by the XSCAPE project on Material Minds.
Avel GUÉNIN
avelguenin.github.io
Reposted by Fluffy Cyborg, IInd of the name
socio-steve.bsky.social
An instructive structure where people feel compelled to good things for selfish reasons is a good incentive structure.
fluffycyborgii.bsky.social
So your argument is that there is functionally no distinction between "military intelligence" and "SD intelligence" ?
fluffycyborgii.bsky.social
Don't you think that it could be the SD private intelligence running ops and tipping the military in exchange for protection ?
Reposted by Fluffy Cyborg, IInd of the name
jmkorhonen.fi
Btw since writing this rather bland statement about Finnish military #intelligence “perhaps helping” deep penetration operations into Soviet Union, I found another history that argues pretty convincingly that the military intelligence did not just help. Likely, they effectively ran some operations.
jmkorhonen.fi
It was in cahoots with Social Democrats, including Nordic sister parties (which also ran their own intel ops) and with Western services, particularly CIA and SIS/MI6.

Early on, the military intelligence at least tolerated to an extent, perhaps even helped Western operations against the USSR.
fluffycyborgii.bsky.social
I told my grandma's Echo in an act of gratuitous rage against the machine, and it yelled at me continuously for five minutes lol
fluffycyborgii.bsky.social
True, "sense of justice" smuggles in the assumption that "justice" is an objective thing you can sense - rather than a convention or a subjective judgement on the states of the world.

Basically enable the "what I did isn't a crime since I'm not a criminal" line of argumentation
Reposted by Fluffy Cyborg, IInd of the name
jostavodebauge.bsky.social
L'histoire méconnue de la "color blindness" :

« Le juge John Marshall Harlan fut la première personnalité importante à se prononcer légalement en faveur de la "colorblindness".
Reposted by Fluffy Cyborg, IInd of the name
dystopiabreaker.xyz
the first thing i did when i got access to gpt3 in 2020 was generate socratic dialogs between von neumann, turing, gödel, nash, and newton
fluffycyborgii.bsky.social
Maintenance in a war environment would require like 6 engineers per robosoldier. I don't see a realistic use case any time soon !
fluffycyborgii.bsky.social
That's something quite incommensurate to what we call "AI" or "robots". And you'd as well just hire a guy, his mother did the work of growing them for you !
fluffycyborgii.bsky.social
Then we come back to that : bsky.app/profile/fluf...

If multi scale competency is indeed critical to life-mike open ended learning behaviour, the problem we have to solve to "grow" it on defined problems sum up to domestication on top of development
fluffycyborgii.bsky.social
Yeah, biological competency relies on a multi-scale organisation that functionally acts as an adaptive "computational" architecture on open-ended problems

Reproducing that in silica is *hard*, harder than (even elaborate) statistical inference on well defined sets of data
segyges.bsky.social
i actually think the paradox has gotten weird. e.g. the robot dogs are actually very very small networks, and clearly work ~somewhat. my best guess is that it's a data problem, it is simply very hard to construct a training objective in silica that resembles "go up some stairs"
fluffycyborgii.bsky.social
I thought that drones in Star Wars were stupid. But what if logistical reasons required their equipment to be interoperable with clones'
fluffycyborgii.bsky.social
And also, most importantly, will maintain inter operability for the foreseeable future at no added effort or complexity
fluffycyborgii.bsky.social
A coherent use case would be "whatever the fuck we don't know how to deal with until we implemented serious automation"

This requires much stronger AI than we have, or will have during my life, and also remains almost certainly more expensive than paying a guy to do it

bsky.app/profile/segy...
segyges.bsky.social
i would expect that you would deploy humanoid robots, if you had viable ones, as your stopgap catch-all to replace humans directly, and then you'd replace them with more specialized robots in turn whenever that was feasible
fluffycyborgii.bsky.social
Well what if you want to feel powerful by yelling at your anthropomorphic butler robot
Reposted by Fluffy Cyborg, IInd of the name
dieworkwear.bsky.social
a high armhole allows for freer and more comfortable movement
Joey Mannarino in a dark brown tailored jacket. The low armhole is causing his jacket to lift. Someone in an inflatable frog costume. The high armhole allows them to raise their arm without disturbing the rest of the garment
fluffycyborgii.bsky.social
What does arguing about the proper measure of "laws of behaviour", when exactly zero evidence supports the existence of such laws ?
fluffycyborgii.bsky.social
It is my unwavering belief that >80% of the methodological issues that empirical sciences have are downstream of epistemological issues.
devezer.bsky.social
It's unfortunate that "p-hacking" and "replication crisis" have become lazy heuristics to call out whatever we don't like in/about/related to science without having to engage with the content or dissect what is wrong that needs correcting.
Reposted by Fluffy Cyborg, IInd of the name
fluffycyborgii.bsky.social
I think if you could pay for the maintenance of such a robot, you could pay a literal army of hospitality services

bsky.app/profile/fluf...
fluffycyborgii.bsky.social
Humanoids just have too many actioners to maintain. Untill we have something like synthetic self-repairing muscle, I don't see it being more than demo