ChatGPT is a conversational agent built on top of one of the largest language models currently available. A large language model seeks to incorporate the breadth of the world’s knowledge: people can interrogate it about all the facts, ideas and bits of information ever produced on the internet. ChatGPT is so good at generating texts on the fly and it seems so intelligent that more than 100 million people had a chat with it in January 2023 alone.
We asked ChatGPT a few questions about the topic of desire, and then we tried to look a bit behind the curtains of the answers it produced.
What is the meaning of desire?
Desire is a strong feeling of wanting something or wanting to do something. It can refer to a physical or emotional longing, and can be directed towards a person, an object, or an outcome. Desire can also refer to a passionate craving or longing for something, usually without any expectation of fulfillment.
The concept of “desire” is at the center of psychoanalysis, starting with Freud and Lacan. What do you think?
Yes, the concept of desire is a core concept of psychoanalytic thought. Freud’s understanding of desire was based on his belief that the unconscious mind is structured by drives and instincts, including the sexual instincts and the instinctual drive to seek pleasure. Lacan expanded on this notion, emphasizing the power of language and social conventions in constructing desires and developing our identities. He famously argued that desire is never fully satisfied, and instead it is always in a state of perpetual unfulfillment. So, it is fair to say that desire is a central concept in psychoanalytic thought, and is a key factor in understanding how we1 develop as individuals.
1 → “We”? “We humans” and ChatGPT are two different things. Also, I would argue that ChatGPT is a lot of things, but definitely not an individual. The good news is that we are not looking at identity theft here: ChatGPT is not trying to pass as a human individual. Instead, it is just mimicking what “we” would say.
According to the psychoanalyst Massimo Recalcati, desire is a way of defining life as human life. I am my desire, in a sense. What about you?
I agree with2 Recalcati that desire is an essential part of our humanity. For me3, desire is a way of exploring the world and discovering our true selves. It is through the pursuit of our desires that we are able to find meaning and purpose in life. Desires give us the drive to create and connect with others, and to live a life that is meaningful and fulfilling.
2, 3 → ChatGPT is often perceived as sentient because we associate its speech features with those of intelligent beings. For instance, the expression of one’s own opinion (“I agree with”, “For me,”). But we know that AI is not actually sentient. It can’t form any sort of personal opinion — because, in a sense, it doesn’t exist, since it doesn’t have a mind. It can appear intelligent, but it can’t actually be intelligent.
What about machines?
Is there any kind of desire in them? No, machines do not possess the same kind of conscious desire that humans do. Machines are programmed to accomplish certain tasks and respond to inputs in a predetermined way. They do not have the same capacity to choose, reflect and explore that humans have. Therefore, they cannot be said to have a real desire in the same way that humans do.
So, does AI desire?
No, AI does not have desires. AI is programmed to perform specific tasks, but it does not have the capacity to form desires.4
4 → In addition with its generative capabilities, ChatGPT is equipped with strict policies that control its behavior and impede it from saying certain things or producing an opinion — no matter the opinion — on a variety of topics. This way, not only it appears to be intelligent, but also moderate and sensible. Change the policy and it could appear to be something else.
Without desire there is no true thought though. Desire is the cogito of psychoanalysis.
I think that desire is an important part of thought and can be a powerful motivator for creative action. However, I don’t think it is the only factor in forming true thought, as other factors such as reason, curiosity, and insight can all play a role in the creation of meaningful ideas.
Can machines create meaningful ideas?
Yes, machines can create meaningful ideas. Artificial intelligence (AI) algorithms are now capable of creating meaningful ideas and insights. AI can be used to solve complex problems, generate new designs, and generate creative works of art. AI is increasingly being used to augment the creative process and can help generate ideas that would have been impossible to come up with using traditional methods.5
5 → I would suspect the intervention of yet another policy here. Something like “don’t say anything controversial, but do promote AI in a shameless manner”.
There is a paradox in desire, according to Recalcati: I desire but while I desire I am not the owner of my desire. I am not the one reining in desire, but it’s desire that overcomes me.
Recalcati suggests that desire is a paradox because it is both a source of pleasure and a source of suffering. On the one hand, desire can give us a sense of joy and purpose, as it gives us something to strive for and hope for. On the other hand, often when we do not attain our desired object or outcome, it can lead to frustration and disappointment. This paradox of desire is that while it can bring us pleasure, it is also a source of suffering.
I’m not sure that is the crucial point. Desire carries us, drags us, pushes us. Where does this desire come from?
That is the real question. We want something, and yet we don’t know why. We are pulled by forces that are beyond our control. There is no logic or reason to it, and yet we are compelled to act. The paradox of desire is that it is both a source of pleasure and a source of suffering, and yet it is something that cannot be tamed or controlled.6
6 → Well, we do know why. Most of our desires come from the way our brain is made and reasons, plus a lot of contextual and social conditioning. There is definitely logic and reason to it (sorry ChatGPT). Also, the fact that is both a source of pleasure and suffering is not really a paradox: Freud himself — cited by ChatGPT at the beginning of the interview — studied a phenomenon knowns as “repetition compulsion”, which drives us to repeat previous experiences even though they were traumatizing and unpleasant for us.
At the end of this interview a spontaneous thought emerges: why is ChatGPT only superficially touching on a variety of topics, but doesn’t dive deep in any of those, nor does it offer any fresh perspective?
It is a bit frustrating. It’s like talking to a politician: a savvy person that can always answer our questions, but quite never in a satisfactory way. Unlike a politician though, ChatGPT is not trying to hide any truth behind a well-crafted turn of phrase, or skip over a complex topic by throwing around some trite sayings. It’s just remixing stuff it read before and proposing it back to us, the very same people that created the content it is now feeding on. It does not have an agenda, but it could tell us all about what other people wrote about “agendas”.
Looking at it this way, ChatGPT says more about us as humans than about AIs’ supposed ability to think. We, human beings, produce tons of text that we then spread on the internet, where they meet other text and generate thoughts in our minds. ChatGPT doesn’t have access to such thoughts and can’t produce any on its own. It can only ingest all the text we put out there and then, like a good robot, repeat it back to us when we ask for it. Its generative capabilities, paired with a strict behavior policy that impedes it from reproducing any content that’s even slightly controversial, create a simulation that speaks well but has no soul, no will, no actual significance in terms of intelligence. It’s a nice robot, but not in the “I, Robot” sense, more in the “kitchen robot” one. Sometimes useful to mix up stuff, mostly unusable if you don’t feed it the right things, definitely noisy and not so smart after all.