Post B5ohWmdGd7zyomDSym by futurebird@sauropods.win
(DIR) More posts by futurebird@sauropods.win
(DIR) Post #B5ogzea3XsymI0uto8 by futurebird@sauropods.win
2026-04-30T10:30:48Z
0 likes, 0 repeats
This is not what the people who say the word AI right now are doing at all. But let's consider a one-to-one physics based simulation of the brain of a cat. It couldn't run in real time like a living cat. It would be very slow, but you could give it reasonable inputs and get responses similar to how a cat might react (the only reason not to do a human mind is size, this is going to take so much computing to do very little)Is it conscious like a cat?I lean on a hard YES.
(DIR) Post #B5oh51axPCYx7Etowa by futurebird@sauropods.win
2026-04-30T10:31:45Z
0 likes, 0 repeats
And I will follow that with: destroying the simulation carries similar moral weight to killing a cat.
(DIR) Post #B5ohBY9fGINJpsFF8C by statsguy@mas.to
2026-04-30T10:32:55Z
0 likes, 0 repeats
@futurebird Oh, that's way too easy. You create a computer simulation where you put in your desired inputs and then the computer completely ignores you and does something completely different.Voila, a perfect cat simulation.
(DIR) Post #B5ohERMywgO1f0quEy by davep@infosec.exchange
2026-04-30T10:33:27Z
0 likes, 0 repeats
@futurebird Id be interested to read why you lean on a hard yes.
(DIR) Post #B5ohJxZucdwL09X3sO by futurebird@sauropods.win
2026-04-30T10:34:28Z
0 likes, 0 repeats
OK, making a simulation like that where you model every bit of matter in the brain of a cat is probably beyond our understanding of matter and more to the point it's slow and will take too many computers to run. Simulations are supposed to be smaller simplified versions of the real things they represent. You could probably simplify the model. Remove bits and replace them with simplified modules that have the same results 99.9 percent of the time.
(DIR) Post #B5ohWmdGd7zyomDSym by futurebird@sauropods.win
2026-04-30T10:36:47Z
0 likes, 0 repeats
@davep Well it's kind of the same reason that it's wrong to cut down the largest and oldest tree in the forest. It's the complexity, the beauty of such a system, it has intrinsic value. Why is consciousness something that can be experienced (or why does everyone report this is true?)Maybe such a simulation would NOT report that it is alive. That seems unlikely.
(DIR) Post #B5ohly7yAogQoUolyi by futurebird@sauropods.win
2026-04-30T10:39:30Z
0 likes, 0 repeats
@welbog Greg is on here! @gregkh dude makes some really good posts about interesting math.
(DIR) Post #B5ohtUcvjwL1dSt4vw by pug50@toot.community
2026-04-30T10:40:51Z
0 likes, 0 repeats
@futurebird It's sad to remember that this sort of interesting/terrifying stuff WAS what AI research was a while back.Although, now I do want to see what a cat controlled by fancy autocorrect would be like... Presumably far more eager to please? Did I just invent "dog"? đ
(DIR) Post #B5oi1bAdILZoFMW5A0 by jmax@mastodon.social
2026-04-30T10:42:19Z
0 likes, 0 repeats
@futurebird Yeahbut. The classic moral quandary of killing is, in Gandalf's words: "Can you give life to those who deserve it? Then do not be so quick to take it."By your scenario, we can bring the cat back. I think that alters the balance. I haven't figured out how, to my satisfaction, but I think it matters.
(DIR) Post #B5oi9wx4CGVm7I9Yvo by semitones@tiny.tilde.website
2026-04-30T10:43:49Z
0 likes, 0 repeats
@futurebird This is a tangent, but we have modeled planarian brains for a while (only 20 neurons!) and those robots do planarian stuff: avoid light, seek food, etc.. Most people wouldn't blink about turning those off, but somewhere between that and a cat, which is an organism we care more about than a flatworm, is a line... But is the line really all that sensible? What is a pet is culturally determined? Animal AG is pretty recent in human history. Hunting to extinction without realizing?
(DIR) Post #B5oiAe4JtArZqEsShs by futurebird@sauropods.win
2026-04-30T10:43:53Z
0 likes, 0 repeats
@jmax I think that's a great point. However, with a system of that size it's not like there is a little save file you can load again. Destroying it really is destroying something forever that you can't get back due to the size.
(DIR) Post #B5oiFMtp9RYy2XDiWe by futurebird@sauropods.win
2026-04-30T10:44:50Z
0 likes, 0 repeats
@semitones People also don't blink about cutting living planarians in half.
(DIR) Post #B5oiTNULl2WxvAz8y0 by jmax@mastodon.social
2026-04-30T10:47:15Z
0 likes, 1 repeats
@futurebird Yes. It might imply a moral obligation to keep proper backups ("We've been saying that for five thousand years!" yell the librarians).
(DIR) Post #B5oilSw6F5o41radM0 by rubinjoni@mastodon.social
2026-04-30T10:50:36Z
0 likes, 0 repeats
@futurebird Physics based simulation:First, imagine a spherical cat...There are several problems with "simulating a cat brain", both technical and philosophical.Are we simulating the action potentials in an adult cat's neural network / brain? Are we taking gene expression into account? How does a cat's brain work, anyway? #Neuromatch people are looking into this.Or, are we simulating cat behavior? What drives the cat to push the cup of the ledge?Or, are we simulating cat's internal 1/2
(DIR) Post #B5oityBZwOK3GTsw9A by Bumblefish@mastodon.scot
2026-04-30T10:52:09Z
0 likes, 0 repeats
@futurebird I think you mist have a very low opinion on the complexity of a cat brain. But consider, cats have domesticated humans. I don't think we're going to get very far into cat brains until we can reasonably model human brains.No, Pica did not write this toot, but I'm not saying that Pica didn't make me write this toot.
(DIR) Post #B5oj1bzEMQ5UjRklhw by futurebird@sauropods.win
2026-04-30T10:53:34Z
0 likes, 0 repeats
@rubinjoni Are we simulating the action potentials in an adult cat's neural network / brain? Yes. Are we taking gene expression into account?Implicitly, yes. It's a scale model of neurons, of receptors and brain fat, it's a physics model of the object called a "cat brain"The stuff surrounding it would be more simplified. The optic nerve causes neurons to fire the inputs are simplified, but the brain itself is a physical model.
(DIR) Post #B5oj7FgYBn6pIrhnBQ by futurebird@sauropods.win
2026-04-30T10:54:35Z
0 likes, 0 repeats
@Bumblefish I have been in trouble with the authorities for so long I no longer flinch at being exposed. But I ask you? Who has the hand and fingers? WHO can operate the can opener???
(DIR) Post #B5ojD8id0GeDR0U6b2 by Bumblefish@mastodon.scot
2026-04-30T10:55:37Z
0 likes, 0 repeats
@futurebird Why worry about opening the can when the human will gladly do it for you, while showering you with praise for your cuteness?
(DIR) Post #B5ojS41kjEZ3EBXwKO by Bumblefish@mastodon.scot
2026-04-30T10:56:45Z
0 likes, 0 repeats
@futurebird I mean after all, they're the idiots who put fish in tins to start with. What kind of intelligent creature does this?
(DIR) Post #B5ojS5KvrPV5HxWlrU by futurebird@sauropods.win
2026-04-30T10:58:20Z
0 likes, 0 repeats
@Bumblefish I'm really no buying that you aren't yet another #PicaTheCat proxy account. Don't you think that she has enough power???
(DIR) Post #B5ojijaXyNBelEloTg by semitones@tiny.tilde.website
2026-04-30T11:01:08Z
0 likes, 0 repeats
@futurebird Right. I dont know about cats, but some people don't blink about cutting guinea pigs in half, or dogs. The line is somewhere. Who is to say it is not between a simulation of a cat, and a cat? Is that any less arbitrary than the other options?
(DIR) Post #B5ok85k8jsuoiDLfg8 by paulc@mstdn.social
2026-04-30T11:05:50Z
0 likes, 0 repeats
@futurebird If you give it a robotic arm does it knock everything off the table that is in reach?
(DIR) Post #B5okDddIALpvGan3lQ by futurebird@sauropods.win
2026-04-30T11:06:57Z
0 likes, 0 repeats
@paulc Yes.And when people say "agentic" that's what I want to see. That's some real agentic action.
(DIR) Post #B5ooGXevESF9QPC11E by libroraptor@mastodon.nz
2026-04-30T11:52:15Z
0 likes, 0 repeats
@futurebird @paulc I'm aware of a robotics lab in KyĹtĹ that makes AI visual processors and walkers. They use a topological gas neural network to interpret the surroundings (via a camera) for surfaces to step on, and their four-legged robots have a spinal cord processor for basic gait plus a brain cortex for modulating the gait depending on the visual interpretation. Their robots can climb ladders and run across uneven ground while the researchers throw obstacles underfoot and make them stumble. They run all this on an Intel 80486 chip, if I remember rightly (it's been a couple of years since I saw one of their papers). It's not a lot of hardware, and not a huge number of neurons, but their papers and videos give me a very strong feeling that these things exercise some kind of autonomous agency.
(DIR) Post #B5op6TCZ38pMeugprk by llewelly@sauropods.win
2026-04-30T12:01:39Z
0 likes, 1 repeats
@futurebird I am convinced the machine learning industry is: (a) heading in entirely the wrong direction for accurate simulations of brains, and (b) in monstrously dishonest and dangerous denial of how much more biology, neurology, and yes, even basic anatomy needs to be learned in order to do such simulations accurately. And there's the fact that brains rely heavily on and interact heavily with all the other organs of the body, and those would also need to be simulated.
(DIR) Post #B5otDX5FPlNQXgtHzk by IngaLovinde@embracing.space
2026-04-30T12:47:43Z
0 likes, 0 repeats
@futurebird @jmax but even if you could back it up, this opens an entire can of philosophical and ethical worms. Good thing there are philosophers working on these worms, and good thing we're nowhere near stimulating a cat and hopefully won't be for a few more decades.
(DIR) Post #B5otLOOva1UBvkqNma by IngaLovinde@embracing.space
2026-04-30T12:49:09Z
0 likes, 0 repeats
@futurebird @welbog @gregkh that's Linux Greg, not hard science fiction Greg! Two very different Gregs
(DIR) Post #B5oznAvXruz5yWjPma by llewelly@sauropods.win
2026-04-30T14:01:26Z
0 likes, 0 repeats
@futurebird given how easily most people are convinced of the "consciousness" of current tech LLM garbage, that kind of thinking has a risk of leading to a future where the rich build huge datacenters to run millions of LLMs, because LLMs have voting rights and legal proections, and vote exactly how the rich tell them to.
(DIR) Post #B5pGyMAu5YfzGWph0i by DavidM_yeg@beige.party
2026-04-30T17:13:53Z
0 likes, 0 repeats
@futurebird If you create a perfect simulation of a cat, you have created a cat. However, I donât believe that is possible, not do I think it will ever be possible⌠I think physics imposes hard limits on our ability to fully know enough to do that. If on the other hand, you have created a cat-like software object, you have created something possibly interesting and unique, but not a cat.imho
(DIR) Post #B5pICNpGqYLeOWxyl6 by JosephMeyer@c.im
2026-04-30T17:27:40Z
0 likes, 0 repeats
@futurebird âIs it conscious like a cat?â If consciousness is a âuseful illusion for controlling the bodyâ or âthe brainâs user-interface for itselfâ as argued by Daniel Dennett with whom I agreed on this topic, then I agree that modeled behaviors would look to an observer like those of a cat and the cat would seem to us humans and to itself like it has consciousness, just like we seem to ourselves and others to have what we call consciousness.
(DIR) Post #B5phvsgjRhe3zulRcu by ophis@brain.worm.pink
2026-04-30T22:15:30.815281Z
0 likes, 1 repeats
@llewelly @futurebird recently i was trying to code a videogame monster room navigation system based on the silverfish in my bathroomand getting frustrated why i needed so many additional checks when irl you could "just" get instant pressure feedback from the antennaeand remembering that to get that in the game i'd need to model two semi-flexible horizontal sticks constantly colliding with the environment in a predictable way, moved by an actor that is always capable of correcting course to prevent either stick from getting caught in the level geometrywhich... would require a lot more checks than i would want to do in this game