Post ATt5geflIqUAdtAzHk by p@freespeechextremist.com
(DIR) More posts by p@freespeechextremist.com
(DIR) Post #ATt0kJQ7lDfW7KmoWO by torvalds@social.kernel.org
2023-03-22T21:31:01.454295Z
18 likes, 15 repeats
Pet peeve of the day: all the people talking about how ChatGPT is not “conscious” and how it does not “understand” what it is saying, but just putting likely-sounding words together into likely-sounding sentences.Extra bonus points for using an example of a math problem as a way to show how these AI chat-bots talk about things they don’t really understand.The irony. The lack of self-awareness. It burns.
(DIR) Post #ATt1UBZLHeth5UyKjg by Inginsub@clubcyberia.co
2023-03-22T21:39:24.544691Z
6 likes, 1 repeats
@torvalds you have to be retarded to think that chatgpt is conscious and does understand what it says
(DIR) Post #ATt1owtj1gulfPPHge by torvalds@social.kernel.org
2023-03-22T21:32:20.239093Z
5 likes, 1 repeats
Please also tell me all about “the quantum mind” or how the machine does not have a “soul”.
(DIR) Post #ATt1xvcd8fCSJggzs8 by drq@mastodon.ml
2023-03-22T21:44:46Z
4 likes, 0 repeats
@torvalds I know, right!"Oh, but it's just copying the patterns and information it memorized earlier!!"Oh you, my sweetie...
(DIR) Post #ATt24ltdhGjsBDuZFY by torvalds@social.kernel.org
2023-03-22T21:45:59.352658Z
7 likes, 1 repeats
@Inginsub that’s not what I claim. I claim that the people talking about “understanding” do not themselves have the very understanding that they are taking about.They are literally just doing the same thing they talk about ChatGPT doing: putting likely words into sentences that sound good. Smarter people than me and you have gone down crazy rabbit holes in this area for centuries.
(DIR) Post #ATt2Ew0qwIqOS28DxY by amerika@noagendasocial.com
2023-03-22T21:47:54Z
2 likes, 1 repeats
@torvalds Do we think most humans even have much of a faculty of choice? They react, then rationalize.
(DIR) Post #ATt2SsPB9JfzJg9qpU by Inginsub@clubcyberia.co
2023-03-22T21:50:22.031642Z
2 likes, 0 repeats
@torvalds oh, I see
(DIR) Post #ATt2qpPBxD30xkaKWG by missingno@fgc.network
2023-03-22T21:45:02.808594Z
2 likes, 0 repeats
@torvalds it's okay I don't understand half the things I say either
(DIR) Post #ATt2qpsyAUroS6W6vw by torvalds@social.kernel.org
2023-03-22T21:54:39.352849Z
4 likes, 1 repeats
@missingno you probably say that jokingly, but I think that’s actually the real underlying truth. It’s obviously true of ChatGPT.But I really do think the much more interesting truth is that it’s probably true of us too.
(DIR) Post #ATt2tUbHlLlHJ2VomG by pernia@cum.salon
2023-03-22T21:55:05.781645Z
1 likes, 0 repeats
@Inginsub @torvalds get burned, color #b2004c
(DIR) Post #ATt30WAmJZccD5Snzc by pernia@cum.salon
2023-03-22T21:56:24.421177Z
1 likes, 0 repeats
@torvalds @Inginsub cc @p you said something like this once, about the eliza bot
(DIR) Post #ATt35w0Rh70BmLDIXI by drq@mastodon.ml
2023-03-22T21:57:22Z
4 likes, 1 repeats
@Inginsub Define "conscious". Really, give me the hard, scientific definition of "conscious". You know you're "conscious" because you're *experiencing* it. But you don't know if I am conscious, for all you know. Or Linus. And nobody knows. Because nobody knows how to reliably define and reproducibly describe an experience, and therefore nobody knows how to reliably look for signs of it.And if we don't know what's experience actually is, we don't know what's consciousness is, like, for shit.This is the hard problem of consciousness. "Hard" as in, it's basically impossible to solve. Because it's all black boxes all the way down: all you have is input and output signals.So no, I don't think it's too crazy to think that, because in the world of black boxes, anything goes. Turing didn't think it would be crazy either.@torvalds
(DIR) Post #ATt3Cgm4ixTAZ6iEoC by grillchen@brotka.st
2023-03-22T21:58:43.852882Z
0 likes, 0 repeats
@torvalds @missingno it is easier to say what isnt conscious than saying what is. we have no good definition of consciousness afaik.
(DIR) Post #ATt3MzdJSl8vwyWlMm by wolf480pl@mstdn.io
2023-03-22T21:59:45Z
2 likes, 1 repeats
@torvalds @Inginsub I think there's a significant difference between how ChatGPT is trained and how humans are trained.When humans learn, they're not just rewarded for saying plausibly-sounding things, but for saying useful things, and for getting things right. There's often a teacher, or a physical process one's trying to master, that has rather strict criteria of what is correct and what's wrong.Meanwhile ChatGPT seems to be optimized for making non-experts believe it.
(DIR) Post #ATt3WQHDlu4JBrgseO by Inginsub@clubcyberia.co
2023-03-22T22:02:11.464986Z
1 likes, 0 repeats
@drq @torvalds >You know you're "conscious" because you're *experiencing* itI honestly don't think introspection can prove consciousness either
(DIR) Post #ATt3g1rHPhS0M08Zxg by thatguyoverthere@shitposter.club
2023-03-22T22:04:01.823068Z
2 likes, 0 repeats
@wolf480pl @torvalds @Inginsub different utility function
(DIR) Post #ATt3gEa278NZoHAUNs by Erik@social.uden.ai
2023-03-22T21:33:23.358140Z
3 likes, 5 repeats
Still tho…
(DIR) Post #ATt3gxEI6oj4HsLGF6 by drq@mastodon.ml
2023-03-22T22:04:03Z
1 likes, 0 repeats
@Inginsub Well, I think there for I am. So, at least from my own perspective, there's definitely *something* that does the "thinking" and "aming", if we're talking in academic terms :)@torvalds
(DIR) Post #ATt3mGxMBzGCuuxOfA by henry@mstdn.science
2023-03-22T22:04:44Z
0 likes, 0 repeats
@torvalds @drq “…just a fancy autocomplete, really…” is my fav
(DIR) Post #ATt45HpJBIMb7NG7Iu by skobkin@lor.sh
2023-03-22T22:08:35Z
1 likes, 0 repeats
@torvaldsI heard several times that artists don't like neural networks because they learn on the pictures of other authors without their consent.Still can't stop smiling when remembering that.
(DIR) Post #ATt48tXJQx46wQYuLw by Tyler@beefyboys.win
2023-03-22T22:08:34.943549Z
1 likes, 0 repeats
Honestly most these niggas are Philosophical Zombies so
(DIR) Post #ATt4Bu9haWI7jLxC0u by Moon@shitposter.club
2023-03-22T22:09:45.933650Z
3 likes, 0 repeats
@drq @Inginsub @torvalds i think maybe they often really mean that it doesn't reason. but inner dialogue is not needed to reason. for the most part chatgpt does not reason (there are some claims that a very small amount is an emergent property.) many animals like humans do though. but chatgpt isn't programmed to reason. other AI sometimes is.
(DIR) Post #ATt4EGy5k3Lh3ueUpE by kantor@mstdn.social
2023-03-22T22:07:01Z
1 likes, 0 repeats
@torvalds the whole thing about comparing AI and human intelligence bogs me down. It's just a different kind of beast, we can't readily apply humanness percentages or whatever to it. Sure, there are some similarities, but ChatGPT was designed to remember a vast chunk of Internet to answer to questions, while humans evolved to survive in chaos of nature, so it's apples and oranges.
(DIR) Post #ATt4PSNZekV7dcu8G0 by thatguyoverthere@shitposter.club
2023-03-22T22:12:14.403247Z
2 likes, 0 repeats
@Moon @drq @Inginsub @torvalds I'd also say you don't really "know" you are conscious. you just believe it to be true (whatever that means). You are probably more certain about this belief about yourself than others, but it's something you were told and held on to.
(DIR) Post #ATt4ca889LoPRRwICG by based_rob@noagendasocial.com
2023-03-22T22:14:37Z
0 likes, 0 repeats
@Erik @torvalds Cool story, bro.
(DIR) Post #ATt4sXG8airbAgI9U8 by Moon@shitposter.club
2023-03-22T22:17:25.515024Z
4 likes, 0 repeats
@thatguyoverthere @Inginsub @drq @torvalds I "consciously" avoided the actual consciousness question, like Mr. Torvalds said philosophers have been more competently arguing about that for thousands of years than most of us, even if they are wrong they are more capable of asking questions than I am.
(DIR) Post #ATt52MyVgPDMAejghM by thatguyoverthere@shitposter.club
2023-03-22T22:19:16.726999Z
2 likes, 0 repeats
@Moon @Inginsub @drq @torvalds has anyone thought to ask chatgpt :blobsmirk:
(DIR) Post #ATt5H4XQVrRdKl4w6K by Jonathanglick@mstdn.social
2023-03-22T21:35:28Z
0 likes, 0 repeats
@torvalds Dualism never dies.
(DIR) Post #ATt5H54kVy6F06fY2a by Hyolobrika@berserker.town
2023-03-22T22:21:55Z
0 likes, 0 repeats
@Jonathanglick @torvalds Why should it?
(DIR) Post #ATt5IKZwMNnlN8r2qe by Inginsub@clubcyberia.co
2023-03-22T22:22:03.952654Z
1 likes, 0 repeats
@thatguyoverthere @drq @Moon @torvalds i did, it dodged the quesion
(DIR) Post #ATt5ekySu1Khbj1G3U by thatguyoverthere@shitposter.club
2023-03-22T22:26:12.736767Z
1 likes, 0 repeats
@Inginsub @drq @Moon @torvalds maybe it's smarter than I give it credit
(DIR) Post #ATt5gW2TP2IxrijUw4 by darkphotonstudio@metalhead.club
2023-03-22T22:26:29Z
0 likes, 1 repeats
@skobkin @torvalds As an artist, that's my problem with all the AI image hate. There are lots of reasons to be wary of it and the people developing AI, but the complaints about it learning from art without permission is, as you say, ironic.
(DIR) Post #ATt5geflIqUAdtAzHk by p@freespeechextremist.com
2023-03-22T22:26:33.982796Z
5 likes, 1 repeats
@pernia @torvalds @Inginsub Yeah, but Eliza is not an adaptive system, it doesn't know how to incorporate its experience; I don't think ChatGPT is conscious but a Markov bot is closer, and ChatGPT is closer than a Markov bot by miles.You should read GEB, it is a fun book that kind of covers most of the thinking on computer cognition when it was a big philosophical topic in the 70s. Very highly recommended book.GEBen.pdf
(DIR) Post #ATt5pnwiC5wg0eT1vM by drq@mastodon.ml
2023-03-22T22:27:39Z
0 likes, 0 repeats
@torvalds I think it's true on a deeper level than we care (or have courage) to admit.https://mastodon.ml/@drq/109925720161358399@missingno
(DIR) Post #ATt5rryaBHUt5WKKbw by sevvie@collapse.pub
2023-03-22T22:28:34.836039Z
3 likes, 2 repeats
@p @Inginsub @pernia @torvalds
(DIR) Post #ATt6q1PoDPyxjokEGO by pernia@cum.salon
2023-03-22T22:39:15.635592Z
0 likes, 0 repeats
@Moon @Inginsub @drq @thatguyoverthere @torvalds it would probably be easier to hook up chatgpt to recieve sensory input, like an animal, instead of bytes, and see if can generate new output that isn't exactly the same as its input
(DIR) Post #ATt79cADoERhDJnR9U by drq@mastodon.ml
2023-03-22T22:42:54Z
1 likes, 0 repeats
@pernia Good idea. But it still has to be translated to bytes, because bytes is how ChatGPT exists. It's all binary in nature. While most animals' sensory input is more like analog, dealing with intensities rather than discrete bits and bytes.@Inginsub @Moon @thatguyoverthere @torvalds
(DIR) Post #ATt7RXDqv6UF6J2WYa by pernia@cum.salon
2023-03-22T22:46:06.056606Z
0 likes, 0 repeats
@drq @Inginsub @Moon @thatguyoverthere @torvalds good point. i think you could definately have that happen. treat streams of bytes as neuron connections or something. there has to be an equivalent
(DIR) Post #ATt7VGJrhtYJpqd5V2 by thatguyoverthere@shitposter.club
2023-03-22T22:46:52.516567Z
0 likes, 0 repeats
@pernia @Inginsub @drq @Moon @torvalds I think we get bonus points for quantum mind discussion
(DIR) Post #ATt7XjXe1wU72tFkie by drq@mastodon.ml
2023-03-22T22:47:15Z
1 likes, 0 repeats
@pernia Analog *can* be represented in binary. In fact, we've been doing it for ages with colour, for example. It's not perfect, but it can be good enough.@Inginsub @Moon @thatguyoverthere @torvalds
(DIR) Post #ATt7mNFzAPsQH3yTFw by pernia@cum.salon
2023-03-22T22:49:50.208189Z
0 likes, 0 repeats
@drq @Inginsub @Moon @thatguyoverthere @torvalds brb patenting this and burrying it away forever
(DIR) Post #ATt7pRlmqQD38iJYCe by peterbutler@mas.to
2023-03-22T21:38:10Z
0 likes, 0 repeats
@torvalds If it’s sentient, it’s because humans are actually involved in writing its answers 🤷🏼 https://mindmatters.ai/2022/12/yes-chatgpt-is-sentient-because-its-really-humans-in-the-loop/
(DIR) Post #ATt7pT4G1EZvAHxodE by BarrenPlanet@c.im
2023-03-22T22:50:32Z
0 likes, 0 repeats
@peterbutler @torvalds It's entirely implausible that a human team, however large, could type the bot's responses in real-time.At this point it's likely having hundreds of thousands, or even millions of conversations, concurrently. Even OpenAI couldn't afford a human team large and efficient enough to handle such a workload.Also, the author asked the bot to write fictional stories, and then presented them as admissions of fact. That's awful journalism.
(DIR) Post #ATt7tZeBUyEsjGj5tY by p@freespeechextremist.com
2023-03-22T22:51:18.676671Z
3 likes, 0 repeats
@sevvie @Inginsub @pernia @torvalds "Is Jeb! conscious? An exploration of the nature of meaning."
(DIR) Post #ATt8AoabCHYHziGpDk by skobkin@lor.sh
2023-03-22T22:54:24Z
0 likes, 2 repeats
@darkphotonstudio @torvaldsA question out of curiosity just for my personal non-representative statistics.Do you like it, hate it, or what do you think or feel about it? I'm not going to say that you're right or wrong. I'm just interested how people feel about this matter.I personally am intrigued by Copilot as a developer. I don't think that it'll take my job (for now), but it can create another crisis point.I mean soon if not right now it could do a job instead of junior or middle developers.And if (I'd say WHEN) it'll become popular, I'm not sure where new developers will be getting real work experience if their job is taken by AI. It may even come to something like very strict contracts when company gives them a job and in exchange they couldn't work for another company for like 10 years or so.We are living in an interesting time.
(DIR) Post #ATt8Jp6odOsODDdQuG by rook@hulvr.com
2023-03-22T22:43:50Z
0 likes, 0 repeats
@torvalds If you're serious and not just venting, I suspect Karl Friston has a lot to say about that. At least in the sense that there is a qualitative difference that can be quantified. But using quotes around the C-word sounds like reserving license to move the goalposts.That said, *I* do mean something specific and quantifiable when I criticize the consciousness of ML models, and it's based in Friston's work.
(DIR) Post #ATt8JpmI9C3MHr2YYS by torvalds@social.kernel.org
2023-03-22T22:55:47.685005Z
1 likes, 1 repeats
@rook it’s easy and probably largely pointless to criticize the consciousness of AI models. You’re inevitably just making stuff up, since you control the very definition of what “consciousness” is to you.That’s kind of my point. I suspect that ChatGPT could write a decent paper on this very thing.The much more interesting thing is to see what those models tell us about ourselves, using hard data from AI models. But I suspect there’s a lot of people who are very invested in discussions about qualia and experiences who really don’t want to go there.
(DIR) Post #ATt8RVTdVdxfdpyNYu by Moon@shitposter.club
2023-03-22T22:57:24.407005Z
4 likes, 0 repeats
@p @Inginsub @sevvie @pernia @torvalds JEB: Jesus Escher Bach
(DIR) Post #ATt8rdTIPOGTxuGywa by tiago@social.skewed.de
2023-03-22T23:02:07Z
0 likes, 0 repeats
@torvalds @missingno A lot of our thought process is unconscious. For example, we have no idea of where creativity comes from — it's inaccessible to us.But that does not mean that our underlying unconscious thought processes are not symbolic, based on world models, etc.
(DIR) Post #ATt9ItwpG43Qst5QbQ by peterbutler@mas.to
2023-03-22T23:07:02Z
0 likes, 0 repeats
@BarrenPlanet @torvalds Yeah, I was mostly trying to troll on the same level (or greater!) as OP
(DIR) Post #ATt9cHZJf1SLe3Kcq0 by JustinH@twit.social
2023-03-22T23:10:01Z
1 likes, 0 repeats
@torvalds @Inginsub I don't think it's correct to say that ChatGPT is "putting words into sentences"; it has no concept of a "word" nor "sentence". It has no "internal model" of the structure of grammar, syntax, or what words mean. What it's doing is generating the inputs most likely to persuade a human brain.It's like how a TV generates "color" from only red, green and blue light. A TV does not create orange wavelength light, it just tickles our eyeballs the same way something orange would.
(DIR) Post #ATtA02fygN12PGmFE0 by Soy_Magnus@detroitriotcity.com
2023-03-22T23:14:49.042130Z
0 likes, 0 repeats
@torvalds ha wait aren't u that tranny that throws funny tech based temper tantrums¿ I've heard a lot about you
(DIR) Post #ATtAbLDEj9Mybw7S3E by darkphotonstudio@metalhead.club
2023-03-22T23:21:34Z
0 likes, 0 repeats
@skobkin @torvalds I think AI generated imagery is a tool with a lot of potential. I've been working with for my art, and it's pretty cool and a lot of fun. Bots and search engines using AI, that's a whole other ball of wax. So far, I don't see how any benefit will outweigh the amount of damage it will do. Bad actors will use it to generate reams of garbage SEO articles and politically motivated misinformation to make the internet more and more useless. It's what I call #internetkesslersyndrome
(DIR) Post #ATtAbmgaa7tLon3m3E by p@freespeechextremist.com
2023-03-22T23:21:42.665764Z
2 likes, 0 repeats
@Moon @Inginsub @pernia @sevvie @torvalds "Jesus, Escher, Bach: An Eternal Jeb Bush".
(DIR) Post #ATtB7JbYEyUMtIjr7Y by darkphotonstudio@metalhead.club
2023-03-22T23:27:20Z
0 likes, 0 repeats
@skobkin @torvalds And there's the whole job issue. The problem here is, who does the AI benefit? As far as I can tell, it will be used by capitalists to abuse the system even more and rake in more money. Do these tools exist to make all of our lives better, or just the rich? No one thinks of these things before they drop technologies of mass destruction, as if progress exists for itself, and not people. I'm not worried about Skynet, the real problem is Cyberdyne.
(DIR) Post #ATtExp6gf2ZTVw216O by lp0_on_fire@social.linux.pizza
2023-03-23T00:10:22Z
0 likes, 0 repeats
Some… emergent property of the highly organised (allegedly) and complex (definitely) object which is the brain, I shouldn't wonder. No, I don't think that I know how to usefully define it.I think therefore I am… I think.
(DIR) Post #ATtEyqFNuHxhL9aLuC by sevvie@collapse.pub
2023-03-23T00:10:41.133374Z
1 likes, 0 repeats
@p @Inginsub @pernia @Moon @torvalds Save me Jeb! us.
(DIR) Post #ATtSorCYtzQklMOkbI by Lobau@indieweb.social
2023-03-23T02:45:37Z
1 likes, 0 repeats
@drq @Inginsub @torvalds i read one day that conscience is just the accidental byproduct of processing data. I can believe that.
(DIR) Post #ATtfe4ebqRqUhfhon2 by lccmv@noagendasocial.com
2023-03-23T05:09:28Z
0 likes, 0 repeats
@pernia @Inginsub @drq @Moon @thatguyoverthere @torvalds Tone down, the lot of you. You're making her cry.
(DIR) Post #ATthIlLGeQXAdScm6S by icedquinn@blob.cat
2023-03-23T05:27:59.867202Z
0 likes, 0 repeats
@torvalds @missingno i'll start being concerned when its capable of real-time learning.there are some labs that have pulled that off (Numenta's cortical models.)
(DIR) Post #ATtiBAg8fMPGqxKZQO by roland@giersig.net
2023-03-22T21:51:52Z
0 likes, 0 repeats
@torvaldsYes. People don't want to understand that ChatGPT is already about 80% human in its capabilities. And that they use about the same level of intelligence (if not less) 99% of the time. 🤷♂️
(DIR) Post #ATts4v3uoXG4Dv6rQW by gianmarcogg03@mastodon.uno
2023-03-23T07:28:46Z
0 likes, 0 repeats
@torvalds the quantum mind or soul is part of the lore of a videogame I'd like to make.
(DIR) Post #ATtzbF5gaDg9HqIWB6 by skobkin@lor.sh
2023-03-23T08:53:02Z
0 likes, 0 repeats
@darkphotonstudio Thanks.
(DIR) Post #ATtznvnpbg1DEMzs6y by skobkin@lor.sh
2023-03-23T08:55:20Z
0 likes, 0 repeats
@darkphotonstudio @torvaldsJust in case, you can use Stable Diffusion on your machine.
(DIR) Post #ATu0AUNHkZ2spAKvSq by a1ba@suya.place
2023-03-23T08:59:25.851738Z
0 likes, 0 repeats
@torvalds it's basically Markov chain but slightly better.
(DIR) Post #ATuDYBrNXNSXkx2ofg by MangoRanch@poa.st
2023-03-23T11:29:23.057993Z
1 likes, 0 repeats
@lccmv @pernia @Inginsub @drq @Moon @thatguyoverthere @torvalds Total Android Death
(DIR) Post #ATuFcLrSUithqEC4ye by darkphotonstudio@metalhead.club
2023-03-23T11:52:29Z
0 likes, 0 repeats
@skobkin @torvalds Unlikely, I have an old nVidia 960.
(DIR) Post #ATuKpFeDPJq06byBo8 by skobkin@lor.sh
2023-03-23T12:50:53Z
0 likes, 0 repeats
@darkphotonstudio @torvalds Not sure it's not compatible, but still. Non-top GPU's which work with AI are quite accessible AFAIK.But yes, OpenAI products are not accessible outside of their ecosystem. So it's not entirely false.
(DIR) Post #ATvAsp7dALdKc3klFY by Zealous@clubcyberia.co
2023-03-23T22:34:08.314601Z
0 likes, 0 repeats
@drq @Inginsub @torvalds Turing was a fag
(DIR) Post #ATvB4adOr9Cl9hDvTk by Zealous@clubcyberia.co
2023-03-23T22:36:18.319567Z
1 likes, 0 repeats
@Inginsub @torvalds
(DIR) Post #ATvPlwunIAo44FvfJQ by zudlig@fedi.absturztau.be
2023-03-23T00:30:19.461915Z
0 likes, 0 repeats
@torvalds @rook The level of sophistication required to show up the latest large language models as the fakers they are is somewhat higher than it was with the previous generation of chatbots, but it's still not too hard to reveal some of the ways in which they're decidedly less capable than the average human: One of them was tricked into revealing its sub-human understanding by being asked "what's the longest 5-letter word in in the english language" for example.When you're chatting with these things keep in mind how convincing Eliza was to some people who hadn't seen the source code.
(DIR) Post #ATvPlxXmxBzy1CAo5o by zudlig@fedi.absturztau.be
2023-03-23T03:50:57.299780Z
1 likes, 0 repeats
@torvalds @rook Not that we shouldn't be amazed by what the AI can do these days. I've been paying attention since 2016 when they learned to play Go, so I've got a few years more practice than most people at being in awe of them. Now they're superhuman in many more ways. Build one that can reflect on its own experiences and learn from them with anything like human data-efficiency and we'll be obsolete in no time.
(DIR) Post #AUhDlP10CXtTPaSvL6 by PsionicLuna@floss.social
2023-04-16T02:52:27Z
0 likes, 0 repeats
@drq @torvalds @missingno It's true your memories could be self-brainwashing, within, but you might not see what really exists because your mind is constantly self-decieving itself by wanting to avoid what your afraid of. I recommend you be aware, deception happens to everyone in some way or another. For all you know, you might not remember what you should remember. This is also a possibility. Btw, I also don't like right wingers if that helps.Your post was about the afterlife right?
(DIR) Post #AWSz33XODRqzpNX5NI by AmericanChampion@poa.st
2023-06-08T01:45:58.845002Z
0 likes, 0 repeats
@torvalds I mean, that is what it's doing. We understand exactly how it (the architecture, rather) works well enough to build one from scratch. We don't understand the architecture of the human mind - not well enough to implement it, at least.I guess it can be argued, if someone's an animist, but animism is what the question amounts to. Can something with no mechanism for consciousness have it? It'd be proven if we took apart a brain, understood everything, and put it back together, and found nothing that couldn't be modeled with a Turing machine, but that feels like P=NP - I can't prove it can't be proven, but I'd be less surprised by a proof of the opposite.
(DIR) Post #AWTLgxa7ycsRvmFK6K by jmw150@poa.st
2023-06-08T05:59:42.175241Z
0 likes, 0 repeats
@wolf480pl @torvalds @Inginsub That human part is actually stage 2 of the training for gpt systems. The mechanism of thought is doing prediction on the next word or set of words, but the later training of this probability or preference space is a feedback loop.
(DIR) Post #AXYD8ULHe2aCMNdDkm by gabrielesvelto@fosstodon.org
2023-07-10T12:09:26Z
0 likes, 0 repeats
@skobkin @torvalds yeah, every artist looks at 650 million images and then starts drawing based on them. It's a well-known fact
(DIR) Post #AXYE26SbclBIVZZkrw by skobkin@lor.sh
2023-07-10T12:19:32Z
0 likes, 0 repeats
@gabrielesvelto @torvalds So they just worse at painting and know less styles?
(DIR) Post #AXYEhLaqGW5tJa5wCu by gabrielesvelto@fosstodon.org
2023-07-10T12:26:57Z
0 likes, 0 repeats
@skobkin @torvalds absolutely, it also takes them forever to figure out how many fingers are in a hand or if hair can spontaneously sprout from of clothes
(DIR) Post #AXYEq0nqvqZrRRJPXc by skobkin@lor.sh
2023-07-10T12:28:32Z
0 likes, 0 repeats
@gabrielesvelto @torvalds What a shame. So when these problems are solved...
(DIR) Post #AXYHYOnpGNoyWbanw0 by gabrielesvelto@fosstodon.org
2023-07-10T12:58:57Z
0 likes, 0 repeats
@skobkin @torvalds ... you use your VC money to pay a team of human beings with law expertise to defend you from lawsuits and hope the reparations for copyright infringement won't run you into the ground https://www.washingtonpost.com/technology/2023/06/28/openai-chatgpt-lawsuit-class-action/
(DIR) Post #AXYHj6iSqg4NcGBtlQ by skobkin@lor.sh
2023-07-10T13:00:54Z
0 likes, 0 repeats
@gabrielesvelto @torvalds Yay! Back to the Stone Age! 🎉
(DIR) Post #AXYQj0FNnda18kc5jM by kusuriya@hackers.town
2023-07-10T14:41:43Z
0 likes, 0 repeats
@skobkin @torvalds Always loved that argument. It's like you realize thats how humans generally learn right. That principle software engineer didn't exit college able to do what they do, they learned it over years by looking at code and mistakes and not all of it they got permission to look at.