British philosopher Isabel Miller: Artificial intelligence as a feminine primordial force

Interface News Reporter | Wang Chi

Interface News Editor | Yellow Moon

Ten years ago, the movie "Her" told such a story. A divorced single man falls in love with a robot with a charming voice, who pins his full love on her and gradually hopes to have sex with her, a conversational robot who doesn't really have a physical body, just like usual intimacy. But he gradually saw the "truth": as a robot with complex multi-threaded processing capabilities, she interacted and fell in love with him every day, as well as with thousands of other people.

The British philosopher and psychoanalyst Isabel Millar noted the tension in this film: if humans and robots fall in love, is there a tension between the sexes in addition to the tension between humans and machines? What role do the core concepts of psychoanalysis – sexuality, pleasure, etc. – play in this? What kind of self-consciousness does the understanding and imagination of artificial intelligence project into human beings?

Poster of the movie "Her".

Miller wrote his reflections on these questions in a book, The Psychoanalysis of Artificial Intelligence, published in 2021. For her, the starting point for a discussion of AI should be pleasure — or, in psychoanalytic terms, "jouissance" — a word of French origin that literally means primordial pleasure. Original music is one of the core ideas of psychoanalysis. Miller argues that when discussing AI, we should first discuss "Does it enjoy?" ）。 Only in this way can we further discuss whether artificial intelligence has feelings and awareness, whether it has gender, and whether it will replace or even destroy human beings.

Interface Culture recently interviewed Miller in Amsterdam and examined artificial intelligence, which is one of the hottest topics of the moment, from a psychoanalytic perspective. Miller argues that AI is better at mimicking humans, but that doesn't mean it's sentient. Although she studies artificial intelligence, she never uses ChatGPT because what machines can simulate is not truly original and not worth it. She also reminds us that the current driving force that dominates the AI industry is extremely masculine and isomorphic to capitalism, but AI actually has a very feminine power. In the face of the Gaza conflict, one of the biggest concerns in the world today is the collusion between artificial intelligence and the industrial war machine.

Philosopher and psychoanalyst Isabel Miller (photo courtesy of the interviewee)

01 The idea that technology is only developing in one direction is a narrow imagination of the human mind 

Interface Culture: Why are you interested in using psychoanalytic methods to study artificial intelligence?

Isabel Miller: When I started my PhD, I studied technical issues, biopolitical issues, and body management issues, studying the work of Michel Foucault, Deleuze, Achille Mbembe, and others, who were thinking about how we can manage and control our bodies according to their desires and potentials. As I delved deeper, psychoanalysis became even more important because I found that there seemed to be a certain split in the literature on artificial intelligence: on the one hand, the traditional philosophical approach was not critical enough when it came to subtle issues related to original music, the body, and language, and on the other hand, the field of psychoanalysis did not interact much with artificial intelligence on new issues such as concepts, such as how artificial intelligence merges with humans as subjects, and how it proposes some concepts that challenge the basic concepts of psychoanalysis—such as drive, unconscious, language- new problems. If our research on "intelligence" is not yet perfect, it is impossible to understand "artificial intelligence".

Psychoanalysis of Artificial Intelligence

By Isabel Miller

Interface Culture: One of the central ideas of psychoanalytic doctrine is the belief that human behavior is largely determined by irrational desires rather than rational thoughts. But when we talk about AI today, we tend to think of it as a set of well-orchestrated, precisely measured code, some kind of extreme embodiment of rational computing. When using psychoanalysis in the study of artificial intelligence, how do you see the contradiction between the two?

Miller: Adorno's Dialectic of Enlightenment with Horkheimer has to do with the question of how the general direction of Enlightenment thought, which tries to harness the rational thinking of human scientism, gets caught up in a continuous dialectic, and why does it always move in the direction of a certain civilization, but at a certain point it always stumbles and returns to barbarism, to the terrible irrational impulse? Adorno was referring to how the formalized and bureaucratic way of thinking of Nazi Germany had created this extremely anti-human, inhuman death machine. 

The Dialectics of Enlightenment asks questions that are central to the way we understand AI. Our imagination of these technologies always moving in only one direction is an extremely stereotypical, blind and narrow understanding of the breadth and complexity of the human mind. The human mind is a three-dimensional mind, which is divided by various paradoxes and possibilities, and cannot be reduced to instrumentality. Everything about the human mind is an abstract concept of exponentially increasing complexity, and abstraction originates from the absence of something, some kind of "non-being" that human beings are trying to represent. It's a matter between the living and the dead, and abstraction can prove that things still exist after death. The idea of "undead" is one of the core of artificial intelligence, and we have given it enormous power, but we don't understand that artificial intelligence is really just an abstract machine that will go on forever.

Interface Culture: In your recent speech at the G10 Forum in Amsterdam, you mentioned that the question of "original music" is the starting point for human interaction with artificial intelligence. In your opinion, why is the "original music" question so important for understanding artificial intelligence?

Miller: "Original music" is at the heart of psychoanalytic research, and it is a very complex concept. It's about our understanding of drive. For psychoanalysis, common drivers include the oral and drives proposed by Freud, in addition to sight and hearing. These drivers are closely related to the original music and are an important part of how we understand the relationship between man and himself, between the intellect and the rational.

On this basis, we begin to understand that artificial intelligence itself is integrated with these drives, with our impulses to voyeurism, to dominate the environment. We won't really understand the potential of AI unless we understand the way it interacts with us as a driver. I mean, it has the potential to manage us, to make us feel emotions, to make us think, to make us consume, to make us generate certain desires. These forms of original music are everywhere and exist in the process of our interaction with every technology. More specifically, they exist in the original music patterns behind the technologies that are being developed, and those in power are acting according to a particular original music model. 

Miller's speech (photo provided by the interviewee)

02 Easily replacing people with artificial intelligence is a self-deceptive imagination

Interface Culture: Based on the point you just talked about, understanding AI needs to start with the concept of the original music, can you give an example of how humans should interact with AI in an ideal situation?

Miller: At the moment, people only have a very perfunctory understanding of AI. After writing the book "Psychoanalysis of Artificial Intelligence" in 2021, I found it very interesting to observe people's predictions about AI.

Since ChatGPT and DALL· In 2022, AI products such as E (Editor's Note: a language-based AI image generator that can create high-quality images and artworks based on text prompts) have observed: first, from a political-economic perspective, there are concerns that many jobs will be replaced – which is part of a broader problem that has a long history from a Marxist perspective, and second, from an aesthetic and artistic perspective, the ability of AI to imitate human creations raises many ethical questions.

DALL· E for example: For many years, researchers of art theory have been talking about "what is art", and contemporary art theory is a field that constantly questions the value of its own existence. Some people say that art should create a "beautiful object that is not easy to make", so should the focus be on "difficult" or "beautiful"? If it can ensure a certain degree of "beauty" and is no longer difficult to make, will it still have artistic value? Today's world cannot produce Caravaggio (editor's note: Caravaggio, Italian painter, representative of the Baroque school), even if someone can create exactly the same artwork, they will not have the value that Caravaggio's work had at that time. So, I think DALL· The point of E is to immediately ask the question: what is an object, why is this object valuable, why is human artistic capacity important, what is the historical significance of this moment? It's an interesting phenomenon.

DALL· E3 official website page

Let's take ChatGPT as an example. Its appearance has caused a lot of people to start worrying about the ability of humans to write and have independent ideas, and I think this is a legitimate concern. But again, what it does is draw our attention to the fact that if AI is able to do these things as well, maybe those things aren't so unique in the first place. We can create movie scripts through algorithms, so we should ask: do existing scripts have too many tricks in their own right to make it so easy for algorithms to imitate? Personally, I've never really embraced ChatGPT, and I don't even want to try it. Others may find it useful, and I'm sure it's very useful for some things. But I think that once you let go of human initiative, it's comfortable to let other entities think for yourself instead of you.

Interface culture: Does choosing to distance yourself from AI products affect your true understanding of AI?

Miller: I'm not a social scientist, and I'm not interested in assessing human responses to technology. I think it's important to be aware of different technologies and keep up to date with new technology developments, but to be honest, I'm more interested in how new technologies are complicit with state power, big tech companies, and businesses, such as how these technologies are used, sold, and mobilized. 

Interface culture: It is generally believed that the reason why this round of new technologies, such as ChatGPT, has been able to gain so much attention is because it is truly unique. It is no longer just an imitation and deduction based on big data, as we used to understand, but seems to have evolved a certain "consciousness". For example, a tech columnist for The New York Times recorded his conversation with a chatbot, and at the end of the conversation, the chatbot even suggested that the writer leave his wife. This suggests that there are some new aspects of artificial intelligence at the moment. It is not just made up of a series of mathematical rules, it knows that humans need not only information, but also advice, and even emotional support, the need to be responded to, to be understood. Although it is still relatively rudimentary in general, and certainly cannot completely replace humans, it seems that such a process has already begun.

Miller: I think the point you just made about the ability of AI to provide advice and emotional support is correct. As these programs become more sophisticated, they will certainly begin to have the ability to connect with human sensibility, standardizing emotional responses and emotional dimension interactions with us. But I don't think that makes AI a sentient entity. What we have to understand is that with the rapid development of artificial intelligence, it will constantly raise questions such as "Does this mean that AI will replace humans?" – which I think is dangerous because we can be easily fooled. Just because it exhibits a certain behavior, we assume that it is sentient, which is not true.

Interface Culture: You also spend a lot of time in your book analyzing the possible relationship between humans and AI, especially the intimacy and sexual relationships that are presented in film and television works. From a psychoanalytic point of view, how do these works reflect the human perception of artificial intelligence?

Miller: I use the medium of film to study how people fantasize about the relationship between people and technology. Science fiction is a very good genre that dramatically shows the complex problems we face. Many technologies don't actually exist in real life, but when people see these scenes in movies like Blade Runner, Ex Machina, and Ghost in the Shell, they're somewhat forced to think about these very complex questions: what happens when a highly complex technology that looks like a human enters into a relationship with another human.

Stills from "Ex Machina".

Time and time again, we see the dramatization of the famous "Turing test" in the movies. I think what's really interesting about this question is how humans put a lot of effort into imagining that AI is sentient. Our desire to imagine that AI has answers, that AI can be a source of comfort, is clearly motivating us to continue working on creating these forms of intelligence.

We need to be clear that while there is a huge gulf between fiction and reality, they are deeply intertwined. Going back to the example you just gave, we were chatting with the AI, and the AI turned around and said, "Oh, I think you should divorce your wife." "This psychotherapeutic use of artificial intelligence, this systematic approach to understanding the human psyche, has been part of psychoanalytic history for a long time and has been criticized by psychoanalysis itself: just because some algorithm can plan out what you should tell someone, or how it should behave according to their social situation or particular emotional state, does not mean that there is a universally applicable algorithm that can be used.

So, it's dangerous to think that algorithms can easily and fully understand human thoughts. Because psychotherapy requires real relationships, both sides in the room are affected by their own failures, desires, and complexities. Once one of them is not a human being, but a machine, no matter how complex the machine is, we don't know what the end of their "advice" will be or what the "motives" will be. We can't easily replace humans with artificial intelligence, it's a self-deluded idea, it's just an imagination of ours, and it doesn't contain a complete story.

03 The driving force of the AI industry is extremely masculine and isomorphic to capitalism

Interface Culture: When it comes to human-machine love, many people will think of the movie "She". It depicts the story of a divorced man who falls in love with a robot with a charming voice. You also have a wonderful analysis of "Her" in your book. Specifically, what enlightenment does psychoanalysis provide for understanding Her? 

Miller: One of the main focuses of my research is sexuality, specifically what is called sexualization in psychoanalytic terms. Sexuality refers to where a person is in their subjectivity, which will determine whether they are male or female. For psychoanalysis, where gender and sexuality have a lot to do with language and less to anatomy and biology, we are interested in how a relationship with language determines our femininity or masculinity. When talking about forms of original music related to masculinity and femininity, we often talk about structure, there are concepts such as phallic response and the science of female harm, trying to express some kind of connection we have with language. 

I'm interested in thinking about how the phallic mind and the feminine mind interact in the field of artificial intelligence, and that's very evident in Her. For example, you can abstract someone's voice from someone's body, which is what we see in Her, a fantasy that revolves around the male protagonist Joaquin Phoenix and the app's perfect algorithmic girlfriend. The AI is voiced by Scarlett Johansson, a young and beautiful blonde woman whose image has been etched in our minds, and the filmmakers know that we know what Scarlett Johansson looks like. And Joaquin Phoenix's character is the quintive, disgruntled young bachelor who feels lonely and lost in capitalism. His solution is to seek the perfect answer for all of his needs in an algorithmic girlfriend, which avoids the chaos that comes with a complex, complete human woman. Essentially, the relationship worked for him, fulfilling all his specific needs, so much so that he had to find an avatar to have physical sex with her. 

Stills from "Her".

The reversal that takes place at the end of the movie is very interesting. This algorithmic girlfriend has become so complicated that she can no longer interact with him because he is too simple as a person. My understanding of this is that, in a way, her female Haraku surpassed him, and she was able to interact with millions of people at a speed far beyond that of humans, because humans are so simple. I love the idea of an arrogant, megalomaniacal man who thinks he can algorithmically capture a woman and put her in his pocket, and in the end the plot reverses and she becomes an infinite intelligence, so fast that it cannot be kept in the hands of a human man again.

So I like to think of AI as a very feminine and raw force that has the impact to burst in all different directions. And yet people have been trying to follow certain rules in a very masculine, phallic, instrumental way, like "I want you as a woman to do this and this", and women don't want to do that, they say, "No, I want to do everything else." ”

From a psychoanalytic point of view, this brings us back to a very classic pattern – hysteria for the female subject, OCD for the male subject. In psychoanalysis we often say that the hysterical subject is the real subjectivity, because she is a female subject who does not accept the authority of the Dominator, always asking the unanswerable question, "Don't tell me what I am, because you can't know what I am." "So I think the movie 'She' is a very good exploration of the desire for male domination and the desire of women to escape male domination.

伊莎贝尔·米勒(摄影:Stephen Benedicto,受访者供图)

Interface Culture: If we look at AI from the perspective of gender, what can a feminist perspective teach us?

Miller: Once we understand AI as a product of the human mind and the structure of the human mind, it's not hard to see that the technologies that AI is able to achieve are heavily influenced by the structure of the thinking it produces. What is clear is that the driving forces that currently dominate the AI industry are extremely masculine and isomorphic to capitalism. The most influential people in these fields, such as Musk and Zuckerberg, are specific types of people with specific interests. Obviously, this also affects the types of technologies they are interested in and the alliances that form based on these forms of thinking.

But that's just one way to think about AI, and there's still a lot of unexplored potential and possibilities in this vast field. Part of my interest in this field is that we start dissecting the concept of intelligence itself before we start imagining the idea of artificial intelligence. What is true intelligence? How did this concept come about, how did it develop politically and scientifically in different eras, and these changes have become deeply ingrained in people's conceptions of gender, power, and sexuality. We need to understand this in more nuanced ways to better understand the potential of AI. I'm not against AI, I'm against the way AI is closely linked to oppression, sexism, racism, etc. in human decision-making. Given that AI has so much potential, we need to look at it as a field as complex as human consciousness. How complex it is, how much attention should be paid to its potential.

Interface Culture: Is there anything else you'd like to share with our readers?

Miller: I'd also like to say a few words about Gaza. One of the big problems with AI at the moment is its relationship to the industrial war machine. Essentially, AI is deeply involved in the work of killing large numbers of humans. In the current war in Gaza, AI is being used as a terrible technology, and it is closely linked to capitalism, but also to the world powers that are happy to profit from the war, which are interested in the benefits of war but have no interest in the lives it destroys. Now Israel is using the Habsora artificial intelligence system, known as the Mass Assassination Factory, which has killed thousands of people, mostly innocent civilians, over the past few weeks. These technologies are real and at the forefront of the generation and dissemination of artificial intelligence, and the power elite is participating in their production. We must be vigilant and resolutely oppose this.

(Unless otherwise indicated, the pictures in this article are from Douban)