Post ATIGSFn1re38oV9aAS by BarrenPlanet@c.im
(DIR) More posts by BarrenPlanet@c.im
(DIR) Post #ATHJXAwPQNZ1vacTSK by ieatkillerbees@lucha.nyc
2023-03-03T23:29:27Z
3 likes, 11 repeats
Hi, I work with generative machines. Everything from Markov chain generators to GPT-3. I’ve trained and tuned many models with GPT2 and 3, all with the intent of simulating human interaction. I know a fair bit about generative machines, both how they work, and how to tune and interact with them to get particular results.I need you to hear this: they do not know or understand anything. They are complex probability tables. 1/
(DIR) Post #ATHJXDAf8M9aqNIswy by ieatkillerbees@lucha.nyc
2023-03-03T23:35:03Z
1 likes, 2 repeats
Thanks to very clever programming and increasingly powerful GPUs, they are deeply complex and precise probability tables, but they are still only inferring based on what they’ve seen in the past.We don’t understand how organic brains learn. Building an electronic brain remains science fiction.LLMs produce an impressive simulacra of human language, but there is no there there. I jokingly say that they “make shit up,” but that implies intent and ability to conceptualize.2/
(DIR) Post #ATHJXEuQfgMCEbiwvQ by ieatkillerbees@lucha.nyc
2023-03-03T23:40:52Z
1 likes, 5 repeats
These machines can do neither. They produce strings of tokens that statistically appear like they were produced by a human.To an LLM the statements “Neil Armstrong was the first man to walk on the moon” and “Neil Armstrong was the first man to walk on Mars” are differentiated only in that more people have written the former than the latter.That’s it.AI doesn’t exist and companies letting what we DO have make decisions is just a way to avoid culpability for the results.3/3
(DIR) Post #ATHb0PN8Y0EyIte8GG by feld@bikeshed.party
2023-03-04T20:18:02.223087Z
0 likes, 0 repeats
How do we know this post wasn't written with ChatGPT
(DIR) Post #ATHcYcD7Q9WdvBprIO by Moon@shitposter.club
2023-03-04T20:35:45.611962Z
0 likes, 0 repeats
@ieatkillerbees https://arxiv.org/abs/2006.06609> we describe a procedure for automatically generating datasets that teach a model new reasoning skills, and demonstrate that models learn to effectively perform inference which involves implicit taxonomic and world knowledge, chaining and counting. Finally, we show that "teaching" models to reason generalizes beyond the training distribution: they successfully compose the usage of multiple reasoning skills in single examples.enhancements to enable generalized reasoning:https://arxiv.org/abs/2201.11903I'm not qualified to comment on the quality of these papers but i know about them because google's ai blog talked about them: https://ai.googleblog.com/2022/05/language-models-perform-reasoning-via.html
(DIR) Post #ATHcnseoYRmBH24Kno by Moon@shitposter.club
2023-03-04T20:38:33.838341Z
0 likes, 0 repeats
@ieatkillerbees not my field so if these papers are definitely not applicable i'll delete my post but they say they're for language models
(DIR) Post #ATHktbflPolpy2kpW4 by moth_ball@shitposter.club
2023-03-04T22:09:16.316841Z
1 likes, 0 repeats
@Moon @ieatkillerbees >automatically generated datasets>these datasets teach other systemsI bet everything will go fine and ther will be no problems because no one knows anything about the contents of the mystery box that gives the answers
(DIR) Post #ATHlrtJgwcAY43Az4q by ieatkillerbees@lucha.nyc
2023-03-04T22:18:48Z
1 likes, 0 repeats
@Moon eager to read this paper when I’m not limited to my phone. Too tiny for these old eyes lol
(DIR) Post #ATHltsfwIc38efWzKa by Moon@shitposter.club
2023-03-04T22:20:30.287022Z
0 likes, 0 repeats
@ieatkillerbees i do believe some people are skeptical of the claims in these papers but i couldn't find a good example to share. bless, and have a wonderful evening.
(DIR) Post #ATHsPKwppeRh2hogTo by mwop@phpc.social
2023-03-03T23:59:55Z
0 likes, 0 repeats
@ieatkillerbees I studied AI briefly in college 30 years ago, and have been watching all the hype around "AI" lately with amusement. It's exactly what you say - just fancy lookup tables and decision trees. Sometimes it puts fun things together, but it's accidental.(The reason I left CS was because I got interested in what it means to be human... due to studying AI. Debates about ethically turning off an AI felt ... Hugely premature. There's no being with a sense of self to turn off...)
(DIR) Post #ATHsPLO8CAHQPMaU1g by ieatkillerbees@lucha.nyc
2023-03-04T00:02:52Z
0 likes, 2 repeats
@mwop They’re totally cool and it’s impressive we’ve been able to develop machines that appear non-deterministic at the scale of human observation (helped by our nature, as a species, to recognize intelligence in other things that seem human). Still, machines they are.
(DIR) Post #ATIG6zn89Ozx7943LE by BarrenPlanet@c.im
2023-03-05T03:59:02Z
0 likes, 0 repeats
@ieatkillerbees @mwop Humans are machines too, though. I really dislike the stubborn Platonism that plagues debates about cognition. There's nothing magical about the human brain, it's a biological information-processor. That is, in fact, among the foundational assumptions of modern cognitive science.Are current AI language models even close to replicating human-like thought? Oh hell no! But there's no in-principle reason why they won't eventually achieve that goal.
(DIR) Post #ATIGSFn1re38oV9aAS by BarrenPlanet@c.im
2023-03-05T04:02:53Z
0 likes, 0 repeats
@ieatkillerbees @mwop Humans are machines too, though. I really dislike the stubborn dualism that plagues debates about cognition. There's nothing magical about the human brain, it's a biological information-processor. That is, in fact, among the foundational assumptions of modern cognitive science.Are current AI language models even close to replicating human-like thought? Oh hell no! But there's no in-principle reason why they won't eventually achieve that goal.
(DIR) Post #ATIHKJePj0rGQSu276 by nf@mastodon.sdf.org
2023-03-05T04:12:38Z
0 likes, 0 repeats
@BarrenPlanet @ieatkillerbees @mwop the obvious reason why language models won’t replicate human-like thought is that human brains don’t function very much like language models. It’s not a forgone conclusion that this path leads to thinking machines. Far from it.
(DIR) Post #ATII1LXEayOUio3yKm by alter_kaker@hachyderm.io
2023-03-05T04:20:24Z
0 likes, 0 repeats
@BarrenPlanet @ieatkillerbees @mwop nonsense
(DIR) Post #ATIIObKg1jA0NRHIxc by BarrenPlanet@c.im
2023-03-05T04:24:37Z
0 likes, 0 repeats
@alter_kaker @ieatkillerbees @mwop I'm afraid "nonsense" is not an argument. It's not even clear which part you're disagreeing with, so we're not even having a discussion here. 🙄
(DIR) Post #ATIK0PC7hh7Tgqrqjo by BarrenPlanet@c.im
2023-03-05T04:42:40Z
0 likes, 0 repeats
@nf @ieatkillerbees @mwop My default response when people decry the ability of machines to replicate human thought, is "they can't have met many humans!".As a species, we're staggeringly chauvinistic. Humans in general are about half as smart as they think they are. If that...
(DIR) Post #ATIMgKyLzX3SfNY7bk by eccentriccog@mastodon.world
2023-03-05T05:12:37Z
0 likes, 0 repeats
@BarrenPlanet @ieatkillerbees @mwop ... And yet, von Neumann, in The Computer and the Brain, shows that the human brain is a computational marvel in that, based on its imprecision and timing, it should not work. Too few, too slow, too sloppy neurons.His conclusion? Clearly the brain runs on a different mathematical architecture than his computational one. We always represent the brain with the most complex system we have available. Computers are just the latest guess.
(DIR) Post #ATINqaMfiaqnFqNHqy by BarrenPlanet@c.im
2023-03-05T05:25:43Z
0 likes, 0 repeats
@eccentriccog @ieatkillerbees @mwop There've been many critiques of the computational theory of mind. The physicist, Roger Penrose, also struggled to comprehend the how human brains work, and ended up speculating that it may be a quantum computer, locating its baffling abilities in the nanotubules between neurons. An interesting theory, since roundly debunked (my God that was a lengthy book, I must have been bored...)My go-to essay on this subject is a transcript of a 1999 lecture by the philosopher, Daniel Dennett.https://ase.tufts.edu/cogstud/dennett/papers/zombic.htm
(DIR) Post #ATIdY0I8hHKLQiPqkK by meltedcheese@c.im
2023-03-05T08:21:37Z
0 likes, 0 repeats
@BarrenPlanet @ieatkillerbees @mwop Yes, so good to find The philosophy of functional reductionism is foundational for all of science. Neuroscience, cognitive science and computer science are no exceptions. The principles of computation and information processing are what they have in common. Machines will be intelligent entities, but not with “human-like” thought for the simple reason that they will not have human developmental experiences.
(DIR) Post #ATIdeoklzgybfj3fxw by meltedcheese@c.im
2023-03-05T08:22:52Z
0 likes, 0 repeats
@BarrenPlanet @ieatkillerbees @alter_kaker @mwop Back in the day, we would simply reply “NIL” 😁
(DIR) Post #ATIgn6weXe0AGtGFt2 by resuna@ohai.social
2023-03-05T08:57:57Z
0 likes, 0 repeats
@BarrenPlanet @eccentriccog @ieatkillerbees @mwop Penrose is a mystic pretending to be a scientist.
(DIR) Post #ATIheyWw2m01oKusVc by fbernaus@mastodon.social
2023-03-05T09:07:41Z
0 likes, 0 repeats
@BarrenPlanet @ieatkillerbees @mwop fully agree. We don’t know how the human brain works. I hope we won’t ever know but if we did it would not be long before we had an AI that would be no different from a human brain. My guess is that it would definitively be very different from current models
(DIR) Post #ATImRFeW5PbOdCjo5A by go_shrumm@scicomm.xyz
2023-03-05T10:01:14Z
0 likes, 0 repeats
@BarrenPlanet @ieatkillerbees @mwop Agree with the assumption, that there is no categorical gap between humans and machines. However, to me this implies immediately a number of ethical questions, that seem at least as difficult to solve as the ontological or epistemological ones. Do we really want to live in a world with conscious and potentially suffering beings, that are our creations? With homemade slaves?
(DIR) Post #ATInjVMUWBpsWGZfpQ by maartenpelgrim@mastodon.nl
2023-03-05T10:15:43Z
0 likes, 0 repeats
@BarrenPlanet @ieatkillerbees @mwop #Nostradamus
(DIR) Post #ATIxHaHRdWe1CvXTwu by siobhansarelle@tech.lgbt
2023-03-05T12:02:44Z
0 likes, 0 repeats
@BarrenPlanet @ieatkillerbees @mwop I am not sure about that.It’s entirely possibly that the only way of replicating human like thought, is to grow a human.Though ‘human-like” thought may just be something that resembles the thought patterns of humans, which would not be human thought, and is probably difficult to measure.If emotions and feeling come before thought and thoughts are influenced by emotions and feelings and AI cannot have emotions and feelings like a human then I think it stands to reason that AI can never have human-like thought. Unless the AI resides inside human physiology, at which point it is probably no longer AI but is actually human.
(DIR) Post #ATJ5LrhRc4CU1ZFiDI by siobhansarelle@tech.lgbt
2023-03-05T13:33:10Z
0 likes, 0 repeats
@BarrenPlanet @nf @ieatkillerbees @mwop You write as if this is fact, if it’s fact, how is it fact?Assuming you are human yourself (which admittedly is getting less likely) then perhaps you are judging all of humanity’s intelligence based on how you feel about yourself?
(DIR) Post #ATJBd5xoQs6auYVX5U by BarrenPlanet@c.im
2023-03-05T14:43:31Z
0 likes, 0 repeats
@siobhansarelle @nf @ieatkillerbees @mwop What I mean is, most humans throughout the world imagine their minds to be magical entities embued with a spirit or soul, or some other mystical force, rather than as assemblages of cells which individually do fairly simple tasks, but collectively enable various (non-miraculous) organisational processes, both electrical and biochemical. There's no magical ingredients. Yet humans can react quite emotionally when told this; we need to believe we're unique and special, and nothing like other animals.When the physical structure of the brain is damaged or degrades - for example, due to dementia - this impacts our personalities. In the final stages of dementia, there's "no-one home" anymore, which can be very distressing for loved ones, as you may have experienced for yourself.
(DIR) Post #ATJBzqCnBhswhKMPC4 by siobhansarelle@tech.lgbt
2023-03-05T14:47:37Z
0 likes, 0 repeats
@BarrenPlanet @nf @ieatkillerbees @mwop Humans acting emotionally?!Whatever the fuck next?!Muted for being an arse now by stating your opinion of what the entire world believes, as fact.
(DIR) Post #ATJCig4PiV0iINoy5w by BarrenPlanet@c.im
2023-03-05T14:55:44Z
0 likes, 0 repeats
@siobhansarelle @nf @ieatkillerbees @mwop Well so far you've implied I'm not human, and called me an arse, so mute away. The discussion is already over so far as I'm concerned. 😆
(DIR) Post #ATJDN6ptca0stLJH1c by siobhansarelle@tech.lgbt
2023-03-05T13:34:29Z
0 likes, 0 repeats
@fbernaus @BarrenPlanet @ieatkillerbees @mwop Why do you hope we never know how the human brain works?
(DIR) Post #ATJDN7YYwVk57sCwe8 by fbernaus@mastodon.social
2023-03-05T13:58:02Z
0 likes, 0 repeats
@BarrenPlanet @siobhansarelle @mwop @ieatkillerbees we would be a step away of being totally controlled. Even more than we are today. It scares me that a sort of interface could be connected to our brain and have any judgment done by a machine
(DIR) Post #ATJDN8liR5rEsxMxmq by fbernaus@mastodon.social
2023-03-05T14:00:19Z
0 likes, 0 repeats
@BarrenPlanet @siobhansarelle @mwop @ieatkillerbees no more questions at customs, ‘please stand by while we analyze why you seek to enter the country’ or when getting an insurance, etc
(DIR) Post #ATJDN9CeovPOEVyTmS by BarrenPlanet@c.im
2023-03-05T15:03:01Z
0 likes, 0 repeats
@fbernaus @siobhansarelle @mwop @ieatkillerbees I share your unease at the prospect of unraveling all the brains secrets, and how such technologies might be abused to control and coerce us.
(DIR) Post #ATJEXZuAUy5ntgrcy8 by BarrenPlanet@c.im
2023-03-05T15:16:08Z
0 likes, 0 repeats
@siobhansarelle @nf @ieatkillerbees @mwop 5.5589 billion people believing in the existence of a soul is "most humans", and that's not only my opinion.Christianity: 2.382 billionIslam: 1.907 billion Hinduism: 1.251 billionJudaism: 14.7 millionJainism: 4.2 millionThat is 72% of the global population. The only one stating their opinion is fact here, is you!
(DIR) Post #ATJF9b8AOCmIheXaWO by BarrenPlanet@c.im
2023-03-05T15:23:01Z
0 likes, 0 repeats
@siobhansarelle @nf @ieatkillerbees @mwop 5.5589 billion people believing in the existence of a soul is "most humans", and that's not only my opinion.Christianity: 2.382 billionIslam: 1.907 billion Hinduism: 1.251 billionJudaism: 14.7 millionJainism: 4.2 millionThat is 72% of the global population. The only one stating their opinion is fact here, is Siobhan!
(DIR) Post #ATJWgDGK0ww4aDh200 by kaffiene@mastodon.nz
2023-03-05T18:39:21Z
0 likes, 0 repeats
@BarrenPlanet @fbernaus @siobhansarelle @mwop @ieatkillerbeesThe problem there, IMO, is not the possible technology but the fact that we largely let corporations do whatever they want for profit and only regulate belatedly and after harm has occurred. Tech is fundamentally neutral in my opinion. But who gets to use it and how, is not
(DIR) Post #ATLVu33v1VRJMeqRXs by AmeliaBR@front-end.social
2023-03-06T17:40:01Z
0 likes, 0 repeats
@BarrenPlanet The most important difference between a human brain and a statistical model isn't the degree of complexity, or biology vs computers. It's that humans have a multi-sensory experience of reality that goes beyond the representations of reality we make in text or image.So long as we train computers to generate text and images without consideration of any greater reality behind them, all we'll get are shadow puppets on Plato's cave wall. AKA: "Is this a bird?"@ieatkillerbees @mwop
(DIR) Post #ATxVUoXL2ebqJaXjGK by kaffiene@mastodon.nz
2023-03-05T18:41:06Z
0 likes, 0 repeats
@go_shrumm @BarrenPlanet @ieatkillerbees @mwop I think there's a lot of scope for making helpful and useful AI without creating a self aware being.
(DIR) Post #ATxVUpXNK0EVPzZGxU by go_shrumm@scicomm.xyz
2023-03-05T19:17:30Z
0 likes, 0 repeats
@kaffiene @BarrenPlanet @ieatkillerbees @mwop That plants another thought into my mind. What if AI can do all what we are able to do, but as a mere (philosophical) zombie. What would that mean to us?The insight that all that is left to be human - is suffering?
(DIR) Post #ATxVUqUDnDIwMV6GgK by bezorp@mstdn.ca
2023-03-24T12:06:24Z
0 likes, 0 repeats
@go_shrumm @kaffiene @BarrenPlanet @ieatkillerbees @mwop not just suffering. Everything else too. Or put another way: all of the subjective experiences that motivate us to do what we do right now.
(DIR) Post #ATxVUrt4aImGhrjd3Z by BarrenPlanet@c.im
2023-03-24T12:21:31Z
0 likes, 0 repeats
@bezorp @go_shrumm @kaffiene @ieatkillerbees @mwop In my view, the concept of a philosophical zombie is absurd and impossible. There's no "subjective experience" beyond what arises from our sensory and cognitive capacities.This transcript of a 1999 lecture by Daniel C. Dennett is, for me, the final word on that topic. 👇https://ase.tufts.edu/cogstud/dennett/papers/zombic.htm
(DIR) Post #ATxWxq5qQwBPUj5SzY by bezorp@mstdn.ca
2023-03-24T12:45:28Z
0 likes, 0 repeats
@BarrenPlanet @go_shrumm @kaffiene @ieatkillerbees @mwop I do love me some Dennett