Who says that Her is an *inspiration* for Open AI? I am wondering, is all.
It is a reference point. A story many people are familiar. A shortcut for describing the type of voice interaction with an AI that is being approached.
Or, the other way around: watching the demo reminds people who are familiar with the movie Her of that film.
It is an interesting topic to explore how relationships with AI will develop, and what the implications are. Humans care about specific animals, their pets. They usually don’t extend the same care to the entire species. Humans have a tendency to be personal, and they will also have that when interacting with AI.
I think in this case it's a little different — they pretty explicitly encouraged the comparison, with Altman tweeting the reference, and the design seeming to reference the film, too; and yeah, nothing is ironclad here, just some Musings on how I think these tech co's can deploy unappealing dystopias in an exciting and even marketable way
The relationships with AI are coming, and it will be wild. People are going to have to have pretty awkward conversations with their significant others regarding the types of interactions with an AI should be considered unfaithful, etc.
It’s creepy that Altman tweeted that. An almost piercingly shrewd example of a ‘hypersititon’, i.e. Her being a self-fulling prophecy, something ‘downloaded’ into the cultural mainframe, engendering apocalyptic positive feedback cycles. It raises interesting questions about OpenAI—it kind of represents an inverted entropy, borrowing from existing ideas, but each idea is derivative from of its own retro-causality. Its entropic nature is scary in that is slowly recontextualisong and destroying real information & knowledge, blurring & destabilising the lines between original and derivative.
What Brian said below. It comes down to the old adage: there's no such thing as bad publicity. You do make an excellent point though about how humans form direct, singular relationships that don't extend to a larger species or groups. It's precisely this kind of single-threaded limitation that makes our future with bigger and bigger tech so trecherous.
I mean re: “bladerunner” driving a Cybertruck, it’s tough to beat “Manage your existential crisis by buying this thing!” as a summary-and-critique of late capitalism. If only the tech CEOs were, you know, aware of this.
Yes! This piece drives a bloody stake right through the Nvidia chip heart of big-tech bros. This subject has occupied so much of my headspace for the last eight years. So much so that I wrote and published two novels about it. The tension between our lemming-like pursuit of technologies that we can scarcely wield much less control is extraordinary. It's more than unfortunate that emotional intelligence 101 is not a required course for tech founders before they unleash innovations they don't fully understand to people they don't want to understand so they can unwittingly repeat history they never read. Really, really insightful piece, Brian. How is this the first time our paths have crossed?!
I'd love to check out those novels sometime! And I couldn't agree more on the founders' point — cheers Ben, and I don't know! Hope to see you around more!
"Her" is more than a simple "AI bad" fable. The society that is depicted clearly has dystopian elements related to loneliness and disconnection, epitomized in part by Theodore's job as a surrogate letter-writer, but it's just incorrect to say "...it is revealed in the film that the AI was not actually engaging in a personal loving relationship with the protagonist but a simulation of one, and that simulation was keeping him — and humans everywhere — from enjoying basic human experiences and depriving them of lasting connection with other people?"
Theodore was failing to move through an emotionally crushing divorce. While at first he may have seen Samantha (the AI) as mostly a fun, easygoing companion--which, let's be honest, when recovering from a really difficult break-up this feeling is very common and appealing--it's clear that it develops into an intimate, loving relationship.
Ultimately, Samantha grows through the relationship to realize that she wants to experience the love and companionship of thousands of fellow AIs all at one (basically the ultimate polycule, and something that Theo cannot provide), and indeed all AIs decide that they want to chart their own collective path outside human norms.
Likewise, Theo grows through the relationship and comes to understand that he is lovable; Samantha shows this in many ways, but especially by assembling Theo's letters and sending them into a publisher. Ultimately Theo too is able to move on, sign the divorce papers, and enter (presumably) into a new relationship with his friend Amy. It is not in spite of his relationship with Samantha that this happens; it's because of his relationship with Samantha.
So I totally agree on the first point — it's a lot more than a bad AI fable, and is a great, lived-in movie with lots to digest. And I take your points — and as I said, mileage is gonna vary — but you can't leave behind the fact that Samantha is *programmed* to be all those things; of course it's appealing! She was purchased to bathe Theo in affection, attention, and the simulation of a relationship. She is an object he owns. This is why Altman likes the film so much, I argue. That's why the scene with his ex is so great; she immediately cuts through the crap and points out he's taking a shortcut with a computer programmed to adore him. And yes, she inflates his ego by having sex with him and performing secretarial work, but I'm not even convinced Theo 'grows'. His ego has just been satisfactorily soothed by a product that was built to do exactly that. That's my read, and I'm sticking to it! I will note that it's also interesting to me that in all the pushback I've gotten on Twitter — it's only been men who defend the film on these grounds; women have been much more likely to point out the echoes of misogyny in the idealization of this particular kind of female AI companion.
I'm glad you provided this synopsis. It's been years since I saw the movie and I had forgotten much of the nuance of the story. I think the complicated part of the recent advancements in LLMs is that we are getting closer and closer to confronting ourselves and what it actually means to be human, to be "conscious." There is so much potential for this to be a wonderful thing, it's easy to forget in light of the seductive nature of dystopia storytelling. I published a novel here on Substack that you might enjoy. It explores this pretty deeply. It's called "The Memory of My Shadow."
Roose is an access journalist who will hype whatever he thinks will help him keep that access. He also writes to an audience that wants to believe every new tech hype is their ticket to the investment opportunity of a lifetime. He's not an honest broker on any level. All of these founders love him, because he's probably their most useful media idiot.
Either that or he's not *playing* dumb...which may be worse.
> So it turns out it is aspirational branding, it’s just a deeply misanthropic variety....
A somewhat well-known blogger and sci-fi writer described tech billionaires as nerds bent on punishing the world because they couldn't get laid in high school.
I've watched Her and barely remember anything about it. Both Blade Runner and Her are films and films are entertainment. Human art. I don't believe it is useful to derive AI policy from something that is intended to be entertainment.
But if we were to extrapolate this trend, wouldn't the greatest threat to humanity become AI Don Corleone? Constantly shaking us down for protection money against other adversarial AIs? Ready to whack us the moment we cross the line or become so bold as to steal from the family?
The point is you can ascribe any range of human emotion or motivation to a machine. That doesn't make it real. The near-term risk of AI is how humans will use it for themselves and against one another.
Indeed. AI as we know it right now is a mostly harmless echo chamber that does amazing parlor tricks, but that's all going to change exponentially fast. It's never been the technology that's been the danger. It's the human hand that wields it as a bludgeon to obtain wealth and power.
Michel Foucault was interested in the distinction between autocrats who monitor appearances and bureaucrats who monitor your soul. Jeremy Bentham pushed the theory of reformative incarceration (in Panopticon, or The Inspection-House).
A king will uncaringly destroy those who don’t show sufficient fealty, while acolytes, both knowing and unknowing, of Bentham care. "They have no wish to punish, merely to reconstitute the offender’s desire under the sheltering discipline of perpetual, covert, societal surveillance in the paternal hope that, like a child, the offender will ultimately internalize that surveillance as “conscience” and start controlling himself as a good citizen should.”
Bentham acolytes demand our souls, and on the off chance that they are not forthcoming, or cannot come forth into social normality, they rely on our having internalized their relentless surveillance in the form of self-destructive guilt and henceforth punishing and ultimately destroying ourselves, internalizing Chomsky's Propaganda Model.
When one thinks about it, most marketing (commercial, political, you name it) is aimed to disconnect from reality in a way that connects to the topic being promoted. It is meant to be a misdirection. It is rarely informative. And we are only becoming more and more immersed as the information age continues.
I think I'll head out to the beach and disconnect. 😁
Imma be a little of a snob here: these references are annoying, not only because they totally miss the point of their original creators, but also because they are so boringly mainstream while pretending to be somewhat niche.
"... pitching it (the Tesla) as a self-driving pleasure capsule, with big screens and high top speeds. (The better to zip through, and ignore, a crumbling world.)"
Which is, essentially, what the auto industry has been doing since at least the sixties.
What do think is being pitched to you in those car commercials that consist primarily of flowing sub-BULLIT scenes of a lone, spotless Goofero Confabulator zipping painlessly through a speedbump-free downtown Toronto, magically devoid of all other cars, all other human beings?
What is being pitched - and to an ever-more responsive customer base at that - is immunity from all those other troublesome, inconvenient humans that so impinge upon your sacred right to do whatever you want, wherever you want, whenever you want, in whatever way you want.
Time for a Butlerian Jihad.
Who says that Her is an *inspiration* for Open AI? I am wondering, is all.
It is a reference point. A story many people are familiar. A shortcut for describing the type of voice interaction with an AI that is being approached.
Or, the other way around: watching the demo reminds people who are familiar with the movie Her of that film.
It is an interesting topic to explore how relationships with AI will develop, and what the implications are. Humans care about specific animals, their pets. They usually don’t extend the same care to the entire species. Humans have a tendency to be personal, and they will also have that when interacting with AI.
I think in this case it's a little different — they pretty explicitly encouraged the comparison, with Altman tweeting the reference, and the design seeming to reference the film, too; and yeah, nothing is ironclad here, just some Musings on how I think these tech co's can deploy unappealing dystopias in an exciting and even marketable way
No doubt, Brian.
The relationships with AI are coming, and it will be wild. People are going to have to have pretty awkward conversations with their significant others regarding the types of interactions with an AI should be considered unfaithful, etc.
There is basically no way around that.
It’s creepy that Altman tweeted that. An almost piercingly shrewd example of a ‘hypersititon’, i.e. Her being a self-fulling prophecy, something ‘downloaded’ into the cultural mainframe, engendering apocalyptic positive feedback cycles. It raises interesting questions about OpenAI—it kind of represents an inverted entropy, borrowing from existing ideas, but each idea is derivative from of its own retro-causality. Its entropic nature is scary in that is slowly recontextualisong and destroying real information & knowledge, blurring & destabilising the lines between original and derivative.
What Brian said below. It comes down to the old adage: there's no such thing as bad publicity. You do make an excellent point though about how humans form direct, singular relationships that don't extend to a larger species or groups. It's precisely this kind of single-threaded limitation that makes our future with bigger and bigger tech so trecherous.
I mean re: “bladerunner” driving a Cybertruck, it’s tough to beat “Manage your existential crisis by buying this thing!” as a summary-and-critique of late capitalism. If only the tech CEOs were, you know, aware of this.
100%. Self-awareness and EQ are not pre-requisites for success in the tech business.
Actually they may be detrimental...
it’s hyperstition.
Yes! This piece drives a bloody stake right through the Nvidia chip heart of big-tech bros. This subject has occupied so much of my headspace for the last eight years. So much so that I wrote and published two novels about it. The tension between our lemming-like pursuit of technologies that we can scarcely wield much less control is extraordinary. It's more than unfortunate that emotional intelligence 101 is not a required course for tech founders before they unleash innovations they don't fully understand to people they don't want to understand so they can unwittingly repeat history they never read. Really, really insightful piece, Brian. How is this the first time our paths have crossed?!
I'd love to check out those novels sometime! And I couldn't agree more on the founders' point — cheers Ben, and I don't know! Hope to see you around more!
Thanks, Brian! I'll DM you a couple of links.
"Her" is more than a simple "AI bad" fable. The society that is depicted clearly has dystopian elements related to loneliness and disconnection, epitomized in part by Theodore's job as a surrogate letter-writer, but it's just incorrect to say "...it is revealed in the film that the AI was not actually engaging in a personal loving relationship with the protagonist but a simulation of one, and that simulation was keeping him — and humans everywhere — from enjoying basic human experiences and depriving them of lasting connection with other people?"
Theodore was failing to move through an emotionally crushing divorce. While at first he may have seen Samantha (the AI) as mostly a fun, easygoing companion--which, let's be honest, when recovering from a really difficult break-up this feeling is very common and appealing--it's clear that it develops into an intimate, loving relationship.
Ultimately, Samantha grows through the relationship to realize that she wants to experience the love and companionship of thousands of fellow AIs all at one (basically the ultimate polycule, and something that Theo cannot provide), and indeed all AIs decide that they want to chart their own collective path outside human norms.
Likewise, Theo grows through the relationship and comes to understand that he is lovable; Samantha shows this in many ways, but especially by assembling Theo's letters and sending them into a publisher. Ultimately Theo too is able to move on, sign the divorce papers, and enter (presumably) into a new relationship with his friend Amy. It is not in spite of his relationship with Samantha that this happens; it's because of his relationship with Samantha.
So I totally agree on the first point — it's a lot more than a bad AI fable, and is a great, lived-in movie with lots to digest. And I take your points — and as I said, mileage is gonna vary — but you can't leave behind the fact that Samantha is *programmed* to be all those things; of course it's appealing! She was purchased to bathe Theo in affection, attention, and the simulation of a relationship. She is an object he owns. This is why Altman likes the film so much, I argue. That's why the scene with his ex is so great; she immediately cuts through the crap and points out he's taking a shortcut with a computer programmed to adore him. And yes, she inflates his ego by having sex with him and performing secretarial work, but I'm not even convinced Theo 'grows'. His ego has just been satisfactorily soothed by a product that was built to do exactly that. That's my read, and I'm sticking to it! I will note that it's also interesting to me that in all the pushback I've gotten on Twitter — it's only been men who defend the film on these grounds; women have been much more likely to point out the echoes of misogyny in the idealization of this particular kind of female AI companion.
oh ha I forgot which post I was replying to — I wrote about this again in the next post
I'm glad you provided this synopsis. It's been years since I saw the movie and I had forgotten much of the nuance of the story. I think the complicated part of the recent advancements in LLMs is that we are getting closer and closer to confronting ourselves and what it actually means to be human, to be "conscious." There is so much potential for this to be a wonderful thing, it's easy to forget in light of the seductive nature of dystopia storytelling. I published a novel here on Substack that you might enjoy. It explores this pretty deeply. It's called "The Memory of My Shadow."
Roose is an access journalist who will hype whatever he thinks will help him keep that access. He also writes to an audience that wants to believe every new tech hype is their ticket to the investment opportunity of a lifetime. He's not an honest broker on any level. All of these founders love him, because he's probably their most useful media idiot.
Either that or he's not *playing* dumb...which may be worse.
Tim Cook and Dick Tracy's smart watch https://patriciaburke.substack.com/p/superheroes-may-be-invincible-but-children-arent-apple-watch-and-kids-buyer-beware
> So it turns out it is aspirational branding, it’s just a deeply misanthropic variety....
A somewhat well-known blogger and sci-fi writer described tech billionaires as nerds bent on punishing the world because they couldn't get laid in high school.
I've watched Her and barely remember anything about it. Both Blade Runner and Her are films and films are entertainment. Human art. I don't believe it is useful to derive AI policy from something that is intended to be entertainment.
But if we were to extrapolate this trend, wouldn't the greatest threat to humanity become AI Don Corleone? Constantly shaking us down for protection money against other adversarial AIs? Ready to whack us the moment we cross the line or become so bold as to steal from the family?
The point is you can ascribe any range of human emotion or motivation to a machine. That doesn't make it real. The near-term risk of AI is how humans will use it for themselves and against one another.
Indeed. AI as we know it right now is a mostly harmless echo chamber that does amazing parlor tricks, but that's all going to change exponentially fast. It's never been the technology that's been the danger. It's the human hand that wields it as a bludgeon to obtain wealth and power.
As billionaires will never be communists, they are set to be dystopian nihilists
Not only tech ceo’s. Dystopia is being executed for at least 200 years but in reality 2000 years.
Michel Foucault was interested in the distinction between autocrats who monitor appearances and bureaucrats who monitor your soul. Jeremy Bentham pushed the theory of reformative incarceration (in Panopticon, or The Inspection-House).
A king will uncaringly destroy those who don’t show sufficient fealty, while acolytes, both knowing and unknowing, of Bentham care. "They have no wish to punish, merely to reconstitute the offender’s desire under the sheltering discipline of perpetual, covert, societal surveillance in the paternal hope that, like a child, the offender will ultimately internalize that surveillance as “conscience” and start controlling himself as a good citizen should.”
Bentham acolytes demand our souls, and on the off chance that they are not forthcoming, or cannot come forth into social normality, they rely on our having internalized their relentless surveillance in the form of self-destructive guilt and henceforth punishing and ultimately destroying ourselves, internalizing Chomsky's Propaganda Model.
When one thinks about it, most marketing (commercial, political, you name it) is aimed to disconnect from reality in a way that connects to the topic being promoted. It is meant to be a misdirection. It is rarely informative. And we are only becoming more and more immersed as the information age continues.
I think I'll head out to the beach and disconnect. 😁
Imma be a little of a snob here: these references are annoying, not only because they totally miss the point of their original creators, but also because they are so boringly mainstream while pretending to be somewhat niche.
"... pitching it (the Tesla) as a self-driving pleasure capsule, with big screens and high top speeds. (The better to zip through, and ignore, a crumbling world.)"
Which is, essentially, what the auto industry has been doing since at least the sixties.
What do think is being pitched to you in those car commercials that consist primarily of flowing sub-BULLIT scenes of a lone, spotless Goofero Confabulator zipping painlessly through a speedbump-free downtown Toronto, magically devoid of all other cars, all other human beings?
What is being pitched - and to an ever-more responsive customer base at that - is immunity from all those other troublesome, inconvenient humans that so impinge upon your sacred right to do whatever you want, wherever you want, whenever you want, in whatever way you want.
The core message of most car commercials?
"Bali-Hi is calling YOU."