AI is forcing the hand of the Demoscene.
category: general [glöplog]
https://www.newsweek.com/2023/07/21/ai-scariest-beast-ever-created-says-sci-fi-writer-bruce-sterling-1809439.html
So it claims monte-carlo search is better than RL to explore this specific (huge) space of boolean functions. Let's even assume they evaluated it carefully and comparison is fair...
Are you not going too far with this? ;)
Quote:
We are talking about the basis of machine’s self-evolution, no less.
Are you not going too far with this? ;)
I will boldly go where no man has gone before. And you?
The "AI" chip design field is definitely growing fast, so go for it!
BTW this 99.99% reminds me of Pentium FDIV bug.
As for end-to-end machine's self-evolution: I will refrain from any further comments, as I don't want to be the one who detracts you. Just please share some updates with us on wild compos!
BTW this 99.99% reminds me of Pentium FDIV bug.
As for end-to-end machine's self-evolution: I will refrain from any further comments, as I don't want to be the one who detracts you. Just please share some updates with us on wild compos!
A.I = Apelike Idiotbox
The future will definitely be wild.
I have spoken.
I have spoken.
That sterling article, well written, and i do like him as an author but this piece ends up being yet another alarmist clickbait about the speed of technology advancement being pottentialy dangerous. Which i guess sells a lot more papers then a reasonable take on the subject. It goes out of it's way to avoid mentioning that what people call AI on this current hype fad is not the sentience that has driven so much of the apocalyptic robot sci-fi lore. Pattern matching does not have sentience. It's not a black box conjuring ways to take over our world, wish people would stop portraying it as such. Particularly ironic that it mentions Frankenstein's monster when the book had a poignant moral point on how society judged the "monster" as dangerous on-sight and never even gave it a chance to fit in, being a book largely regarded as a tale of people hating on technology for the fear of the unknown aspect of it.
Yes there are intellectual property issues that need to be legislated and yes society will have to adapt itself to new technology. Just like it's been doing for thousands of years. Oh the horror of the masked shoggoth.
Yes there are intellectual property issues that need to be legislated and yes society will have to adapt itself to new technology. Just like it's been doing for thousands of years. Oh the horror of the masked shoggoth.
Well said, PS, I mostly agree with all.
I love Bruce Sterling's novels (those ones in the posthuman world of the "mechanists" and "shapers" are great) but sci-fi doomerism has no substantial predictive value. And, to clarify, I actually do believe than sentient general artificial intelligence will exist sooner than most people would think but I find preposterous all these apocalyptic scenarios that so many people like so much.
All that fear of the end of the world is like a red herring that is distracting us from the real problems that the use of AI could cause and from the safety measures that current systems require.
By the way, I'm tired of all this gratuitous hate towards IA use in the art world and videogames (see the pcgamer link posted before by Gargaj) and here, in our scene. I think that the scene spirit should inspire us to use every new creative tool available and explore its limits and the multiple ways that we could find to enhance human expression and ideation. If someone is a graphician and still don't wanna try all this new stuff then he / she is already a living fossil in my opinion.
I love Bruce Sterling's novels (those ones in the posthuman world of the "mechanists" and "shapers" are great) but sci-fi doomerism has no substantial predictive value. And, to clarify, I actually do believe than sentient general artificial intelligence will exist sooner than most people would think but I find preposterous all these apocalyptic scenarios that so many people like so much.
All that fear of the end of the world is like a red herring that is distracting us from the real problems that the use of AI could cause and from the safety measures that current systems require.
By the way, I'm tired of all this gratuitous hate towards IA use in the art world and videogames (see the pcgamer link posted before by Gargaj) and here, in our scene. I think that the scene spirit should inspire us to use every new creative tool available and explore its limits and the multiple ways that we could find to enhance human expression and ideation. If someone is a graphician and still don't wanna try all this new stuff then he / she is already a living fossil in my opinion.
Quote:
By the way, I'm tired of all this gratuitous hate towards IA use in the art world and videogames
That's your prerogative. I'm tired of the torrent of irrelevant generic art - to me it's noise.
Technology has a tendency to outcast members of the society that do not want to submit or don't even want to care, we label those people "fossils", "old fashioned" and/or "outdated".
It's funny until it happens to you and as we never learn from our past mistakes so we just go forth to the next invention; sometimes the previous invention were desastrous, we just brush it off "blah, they don't understand".
It's funny until it happens to you and as we never learn from our past mistakes so we just go forth to the next invention; sometimes the previous invention were desastrous, we just brush it off "blah, they don't understand".
I do not label people, kaneel. Sorry if I did a bad choice of words but you clearly misunderstood me.
In fact, like many people do, I enjoy (and sometimes even practice) a lot of "old-fashioned" art. You know, things like theater, opera, drawing with pencils, mosaics, cave painting...
Not everything in the art world has to run on a computer! I'm merely pointing out that all evolution in art (sometimes, also society) is preceded by some technological change.
For instance, the discovery and use of the perspective laws in painting during Leonardo's time were preceded by the invention of the camera obscura, etc. Nowadays we are about to experience similar changes in art. It depends on us if the demoscene join the first line of art evolution or... becomes another fossil art (in the sense of its immutability, cause only changing things are alive).
In fact, like many people do, I enjoy (and sometimes even practice) a lot of "old-fashioned" art. You know, things like theater, opera, drawing with pencils, mosaics, cave painting...
Not everything in the art world has to run on a computer! I'm merely pointing out that all evolution in art (sometimes, also society) is preceded by some technological change.
For instance, the discovery and use of the perspective laws in painting during Leonardo's time were preceded by the invention of the camera obscura, etc. Nowadays we are about to experience similar changes in art. It depends on us if the demoscene join the first line of art evolution or... becomes another fossil art (in the sense of its immutability, cause only changing things are alive).
Quote:
I'm merely pointing out that all evolution in art (sometimes, also society) is preceded by some technological change.
I think that demoscene has proved a negation of this proposition to be true.
Quote:
Technology has a tendency to outcast members of the society that do not want to submit or don't even want to care
Correction: Society does, not technology. I'm not worried about AI as a technology, I'm very worried about what the current society will use it for.
@la_mettrie: But how could the demoscene appear without the invention of the computer? It was clearly a preceding technological change that made the existence of the demoscene possible.
In the same way, present and future technological changes will make possible future forms of art that are unknown to humanity today.
In the same way, present and future technological changes will make possible future forms of art that are unknown to humanity today.
to reiterate what i said earlier: we should try to separate ai/ml in general from implementation of the current set of LLMs. interesting technology with huge potential across the board - including graphics (DLSS/denoising/material evaluation/compression/list goes on), vs the anarcho-capitalist abuse of all historical human creativity for the profit of a few large entities. that's not a problem with the technology, it's a problem with the people using it.
@Gargaj: Thanks for the correction.
@smash: nailed.
@smash: nailed.
@psenough:
I have trouble understanding which Sterling article you read.
Or, the other way around, I wonder which article I read.
Obviously Someone is misunderstanding something here.
The only 'alarmist clickbait' in the article is the title (duuuuh!). Which is IRONIC, it becomes clear when you read the article itself.
Sterling is mocking tech bros who make their grand over-the top hype AI claims.
He is being very ironical in mocking the artificially raised hype around LLMs.
After all, this is a direct quote from the article that stuck with me: "A Large Language Model is built from complex statistics, so it's a parrot yakking up its slurry of half-stolen words and images."
This is in direct contradiction to your sentence "It (article) goes out of it's way to avoid mentioning that what people call AI on this current hype fad is not the sentience that has driven so much of the apocalyptic robot sci-fi lore" C'mon, man the whole article says exactly that: "This fad is bullshit!". You say "pattern matching does not have sentience" as if to correct the guy who said the thing (LLM) is "a statistic driven parrot". WHAT?!
And all the while, while it is a parrot and basically a glorified autofill, it still manages to be a black box (as neural networks tend to sooner or later become). It seems to me you don't really understand this tech. Or I don't. It seems to me that you think LLM is a classic database-driven, pattern-matching autocorrect. But I digress, never mind that, let's stick with the article.
From the start to the end of article, the guy mocks exactly that which you accuse him of promoting.
Could you please in the name of God and Skynet explain the correlation between Sterling's sentence I quoted and yours.
I have trouble understanding which Sterling article you read.
Or, the other way around, I wonder which article I read.
Obviously Someone is misunderstanding something here.
The only 'alarmist clickbait' in the article is the title (duuuuh!). Which is IRONIC, it becomes clear when you read the article itself.
Sterling is mocking tech bros who make their grand over-the top hype AI claims.
He is being very ironical in mocking the artificially raised hype around LLMs.
After all, this is a direct quote from the article that stuck with me: "A Large Language Model is built from complex statistics, so it's a parrot yakking up its slurry of half-stolen words and images."
This is in direct contradiction to your sentence "It (article) goes out of it's way to avoid mentioning that what people call AI on this current hype fad is not the sentience that has driven so much of the apocalyptic robot sci-fi lore" C'mon, man the whole article says exactly that: "This fad is bullshit!". You say "pattern matching does not have sentience" as if to correct the guy who said the thing (LLM) is "a statistic driven parrot". WHAT?!
And all the while, while it is a parrot and basically a glorified autofill, it still manages to be a black box (as neural networks tend to sooner or later become). It seems to me you don't really understand this tech. Or I don't. It seems to me that you think LLM is a classic database-driven, pattern-matching autocorrect. But I digress, never mind that, let's stick with the article.
From the start to the end of article, the guy mocks exactly that which you accuse him of promoting.
Could you please in the name of God and Skynet explain the correlation between Sterling's sentence I quoted and yours.
Quote:
You say "pattern matching does not have sentience" as if to correct the guy who said the thing (LLM) is "a statistic driven parrot". WHAT?!
Even at the risk of misinterpreting what PS meant, I should point out that I doubt that all of the "anti-doomer" criticism of your post refers exclusively to Sterling's article. Within the AI debate, the term "stochastic parrot" has been used ad nauseam since the appearance of the first GPTs (I'm not referring to ChatGPT but to GPT and GPT-2). I don't think the statement "pattern matching does not have sentence" necessarily implies disagreement with anything Sterling has said about LLMs in his article, but rather that that sentence serves as the conceptual framework for the next one (i.e., " It's not a black box conjuring ways to take over our world, wish people would stop portraying it as such.") which is a direct criticism of the end of the world discourse being defended by people like Eliezer Yudkowsky or, more recently, Max Tegmark.
And as a side note to the other members of this thread (silent or talkative), please understand that some of us may be using these forums in the hope of exchanging ideas with other intelligent beings. Not with the vain goal of winning Byzantine arguments! We must all use intelligence in the service of truth, not vanity. And we should also be grateful for the opportunity to have a common means of communication (language) in order to facilitate the exchange of ideas and, hence, to improve our mental models with which we interpret the world and not use this common tool (language) to practice rhetorical exercises that serve to feed our ego or confuse imaginary "adversaries". Damn.
edit: all of the "anti-doomer" criticism of HIS post
@ham:
I’ll be brief. Here’s what I see. Sterling mocks hyping up of current crop of AI (LLMs). psenough resents (Sterling’s) hyping up of current crop of AI (LLMs). Doesn’t something feel wrong? Sterling doesn’t do the hyping, he MOCKS it!
P.S. It looks to me that both of them (Sterling, psenough) are of the opposite opinion than, for example, you. It looks to me that they think that the current crop of AI is overhyped bullshit. While (it also looks to me) that you deem that the current crop of AI is something revolutionary and ‘something wonderful’ as Dave and/or Hal would say.
I’ll be brief. Here’s what I see. Sterling mocks hyping up of current crop of AI (LLMs). psenough resents (Sterling’s) hyping up of current crop of AI (LLMs). Doesn’t something feel wrong? Sterling doesn’t do the hyping, he MOCKS it!
P.S. It looks to me that both of them (Sterling, psenough) are of the opposite opinion than, for example, you. It looks to me that they think that the current crop of AI is overhyped bullshit. While (it also looks to me) that you deem that the current crop of AI is something revolutionary and ‘something wonderful’ as Dave and/or Hal would say.
Of course I could have misunderstood something.
Aha, sorry, I explained what I think everybody thinks (jeez what an asshole) but failed to state what I think. Well here goes : I think we in fact ARE in grave danger from the current crop of AI, but not because they ARE intelligent/sentient, on the contrary, I think we are in danger from them because they ARE NOT intelligent nor sentient nor will they be any time soon, but the hype will tend make too many people believe that they ARE. And that’s dangerous.
And, yes @ham, as you said, this talk is not meant to be a dumb sparring match, I too (as you) hope nobody reading it got the impression it was.
And, yes @ham, as you said, this talk is not meant to be a dumb sparring match, I too (as you) hope nobody reading it got the impression it was.
the article goes through several points, the most obvious one as you mentioned seems to be the unnecessary spectacle of it all (what he calls the mardi gras), but then the article repeatedly mentions the dangers of memefying something (with some very bad comparisons to mythical doom creatures being revealed, because i guess you can take the author from fantasy but you can't take fantasy from the author) which seems to be warning against it later becoming accepted at face value and the underlying dangers of that (which to me is nonsensical alarmism but he seems to actually believe that beyond the poetic fantasy comparisons). so his position on this new "AI" wave seems to be more then just mocking the fad riders but more significantly pointing out how the whole thing can be dangerous for society, not just as a byproduct of the hype bros talking shit they don't get (which i think we are all tired of regardless of our position on "AI") but also as a new tech itself, warning against possible future scenarios that will shockingly change society as we know it if not carefully and rapidly regulated / monitored.
yes i agree with him society will change
yes i agree with him this an unnecessary spectacle
how is writing one more click bait spectacle opinion piece helping to address the overwhelming number of other click bait spectacle opinion pieces that been riding the hype? just felt like a lot of pot kettle shenanigans to me.
i do like his writing but this article fuels the disinformation surrounding the topic rather then guiding people into any deeper insight imho.
yes i agree with him society will change
yes i agree with him this an unnecessary spectacle
how is writing one more click bait spectacle opinion piece helping to address the overwhelming number of other click bait spectacle opinion pieces that been riding the hype? just felt like a lot of pot kettle shenanigans to me.
i do like his writing but this article fuels the disinformation surrounding the topic rather then guiding people into any deeper insight imho.
Quote:
Sterling mocks hyping up of current crop of AI (LLMs). psenough resents (Sterling’s) hyping up of current crop of AI (LLMs).
Well, unless PS interviews Sterling on his Youtube channel we can't be sure of anything. :]
However, I do think PS was ranting more about the general current hype of LLMs than about whatever that article said about this. But that's just my impression.
Quote:
I think we in fact ARE in grave danger from the current crop of AI, but not because they ARE intelligent/sentient, on the contrary, I think we are in danger from them because they ARE NOT intelligent nor sentient nor will they be any time soon, but the hype will tend make too many people believe that they ARE. And that’s dangerous.
I mostly agree. People should be aware of the many limitations of current LLMs (transformer-based, also known as GPTs). The tendency to anthropomorphize them is all too common and indeed dangerous.
Seems that he posted while I was writing my previous post. :]
In the age of clickbait, I'm afraid that for an article like that to make the front page it needs to contain quite a bit of clickbait.
The comparison with those Lovecraftian creatures is probably taken from a meme that lately circulated in the AI debate about the brittle methods currently used by Microsoft in order to "achieve human alignment" in Bing.
Quote:
how is writing one more click bait spectacle opinion piece helping to address the overwhelming number of other click bait spectacle opinion pieces that been riding the hype? just felt like a lot of pot kettle shenanigans to me.
In the age of clickbait, I'm afraid that for an article like that to make the front page it needs to contain quite a bit of clickbait.
The comparison with those Lovecraftian creatures is probably taken from a meme that lately circulated in the AI debate about the brittle methods currently used by Microsoft in order to "achieve human alignment" in Bing.