Psychiatrist Sam Lieblich has pushed AI boundaries with his project: PHRASER, a new AI entity that capitalises the future capacities of AI technology.
“Residing in a darkened room, PHRASER consists of a meta-algorithm, a fusion of rules and connections trained by texts produced in workshops run by the artist. The work reshapes current events on the internet, spitting out a monologue of strange words and a fluttering of discordant images.” – via photo.org
Will AI provide humans with the utopia we all desire, or will it achieve the very opposite? In this insightful interview, we ask Lieblich about the origins of PHRASER, embracing the accelerating digital and AI space, and discuss the distinction between dreams and reality – if there is one.
What inspired you to create Phraser?
I had the idea for PHRASER when Elon Musk ran a press conference for his tech firm “Neuralink” in November 2020. The world had spent 10 months in their bedrooms, extremely online, and having a much more tenuous link than usual to their body as a social tool. Their body was not at work, their body was not at play, their floating head was in waiting rooms or meeting rooms, they were in the car pretending to be at home, they were in the home pretending to be at work. Almost all of social life was mediated by one algorithm or another, and Elon Musk was showing us the digital devices his team had implanted inside the skulls of some unfortunate pigs who were shown manipulating a rudimentary digital interface with their brains.
In 2019 Mr Musk said he intended Neuralink to be a “campaign to create symbiosis with artificial intelligence”. The symbiosis he refers to here is meant to evoke a particular figure of techno-capitalist eschatology that Mr Musk, Ray Kurzweil, James Lovelock, Vernor Vinge and many others in their tweets and theories have been calling “the technological singularity”. This term “technological singularity” stands as both the name for a historical “event horizon” over which the effects of runaway technological change become too mysterious to guess at, and as a shorthand for the “thing” many of these theorists fantasise will exist beyond that horizon. The “thing” at the end of all this technological change will be, they say, a human-machine hybrid intelligence of a singular identity into which we’re all somehow assimilated. Such that we are as one hyper mind spanning all of future civilisation; a world spanning or galaxy spanning “thing” that is an über-mensch, an all-mother, a womb, a prodigal son, a blessed daughter, a transcendent subjectivity, a utilitarian panacea, and importantly, a justification for the world-consuming effects of late techno-capitalism. One that imagines at the end of it all is a utopia rather than an apocalypse so we can all feel free to keep going.
Such utopian visions of the world that machines will bring about has included also the notion that machines will take it over and be the more enlightened custodians of it since around the time the first coal powered steam engine pumped water out of a mineshaft in England, and this utopianism may have been quite integral to the social justification for industrial capitalism since that time.
PHRASER answers the question: If the human species is to one day become a single world-spanning hyperintelligence, what will it be stressing about? This quotidian focus problematises the utopian notions of the technological singularity, and instead posits, that because whatever “runaway technological change” might arise will conform to the expediencies of capital, so too, will whatever worldwide hyperintelligence one day emerges. It also problematises the eschatological/messianic dimension of these visions of the technological singularity by considering that the world-spanning hyperintelligence—by dint of the COVID-caused diffusion of our bodies into the mainframe and our pre-existing reliance on algorithms for all social and material provisions—has already arrived and we’re already part of it. As we have already seen with our experiments with PHRASER, our collective hive mind worries who thinks what about us, frets over the to-do list, is pissed off that Instagram thinks we are bald, or need a toilet stool, or want to see pictures of our ex at a wedding. These will be the concerns of the technological singularity in the utopia at the end of technology’s history: Petty bullshit blabbed into the aether for all eternity.
How have people reacted to it?
People find it mesmerising. Some people have stood and watched PHRASER producing speech and images for an hour and a half. I think it’s disconcerting to watch an AI say things that sometimes seem quite self-aware, and most of the time seem like a reflection of things we would say ourselves, but just with that AI uncanniness that language algorithms are now famous for.
What even is a meta-algorithm?
That term is used playfully, because I would say, to paraphrase Jacques Lacan, ‘there is no meta-algorithm’. I mean that any algorithm already involves a set of nested instructions and ‘sub-algorithms’ so it’s always algorithms all the way down. PHRASER is a conglomeration of 4 machine learning algorithms (GPT-2, GPT-3, CLIP, VQGAN) that is tied together with my own custom code and another bespoke algorithm that shuttles language between the other algorithms. It’s this ‘algorithm within an algorithm within an algorithm’ vibe that I was riffing on with my producer Anita Spooner and Zara Sigglekow at the gallery when we came up with ‘meta-algorithm’.
It all seems a bit ‘Matrix-y’. What do you say to the doubters who don’t embrace the digital/NFT/AI space?
This work embodies radical doubt about NFTs and AI so I guess what I’d say is go see the work and see if our doubts are the same.
The Matrix engages with so much of philosophy so it’s hard to address the Matrix-y-ness of the work. I suppose the AI of the Matrix has two aspects: the first is its world dominating, militaristic face, that is that automatons develop desire and agency and the capacity to surprise their creators, and then take over the world. That part is on the one hand, an eschatological myth that embodies the fears that capitalists have that the tools they use to automate profit making will slip from their control. On the other hand, it’s a utopian myth because capitalists (and all of us) know that the current phase of techno-capitalism will consume the entire world, after which there will be no consumers to fuel the system. So as with any social paradox of this kind, a myth is required to culturally resolve the paradox; a thinking, feeling, desiring AI is that myth because it gives people, especially capitalists, a messiah on the other side of the apocalypse, a ‘person’ albeit a digital one, that will still exist after another 100 years of accelerating capitalism and will be able to be the steward of the Earth and of all human cultural production in our name.
NFTs are linked in with this because on the one hand, they’re nothing new, just another tool for the abstraction of value that is no important sense different from money or other value bearing legal/social contracts. On the other hand, they have some practical advantages for the acceleration of hyper-financialization, which always happens in a degenerate society, and which is why the majority of NFTs have no artistic merit whatsoever.
The important thing for me was to offer NFTs that have some chance of having some artistic merit (first of all) but also to launch them at the same time as the Non-Fungible Tokens performance I did, during which I made “NFTs” on paper for people. They would sit with me and tell me a story and between us, we’d try and find a little indissoluble kernel to their story, something that we could make into a ‘token’ which trivialises the story somewhat but also makes it stable like an aphorism. Then I’d write that down on these slips that look like somewhere between a certificate and a banknote, and I’d stamp it with a unique number and give that to them. We didn’t do any documentation of those, no photos, they’re all one of one, the experience we had in making each one is ‘non-fungible’ and the slip of paper participants received is a ‘token’. So, in both issuing NFTs and performing ‘NFTs’ I want to point out that the NFT only has to protest so much that it is ‘non-fungible’ precisely because it is the nature of a token to be fungible.
I wanted to make that same point in the work I made for Blindside’s group show ‘The Portrait’ which is also running now as part of PHOTO2022. There I made a work called ‘Fungible Tokens’ in which an AI and I collaborated on imagining what the various emperors, kings, queens, monarchs of all kinds would look like as they dissolve into money.
I am interested in the way that the sovereign, supreme amongst citizens, becomes associated with a supremely interchangeable portrait. That is, for a coin that bears the portrait of the emperor, this one is just as good as that one. The sovereign who makes an attempt to centre their own personage, ends up dissolving themselves ever more fully in the economy, and becoming co-extensive with the form of value, in the process losing their own qualities.
This happened in an abstract sense at first, with the first coinage bearing quite distinct portraits in side-profile, easily allowing citizens to determine which emperor minted the coin. One could say that at this stage the emperor maintained an aesthetic of a singular identity. In later coinage, starting with late antique Byzantine numisma, the portrait shifts to a 3/4 profile, and then eventually to a frontal portrait. This latter view is less suited to representing the distinguishing features of the face and many think it was chosen to literally dissolve the identity of the emperor. The explicit reason given for this was that it was in deference to the superiority of Christ, but many believe it was rather because there were so many usurpations at the time that emperors wanted to be somewhat anonymous and to more seamlessly blend-in with their predecessors. In this way, they would be less likely to be recognised for their crimes and would sneak themselves into an aesthetic chain of legitimacy. The emergence of the blockchain and of Non-Fungible Token mirrors this same process, from state-sanctioned currency (which was eventually entirely co-opted by powerful corporate and financial entities) to totally decentralised currency, anonymous and fungible despite explicit claims of non-fungibility.
I am very interested in the way that the One becomes dissoluble in the Many when it is made into a token. One can only imagine that in a post-panama-papers world, capital is looking for ever darker hiding places for its crimes as the malignancy of the value form threatens to consume the entire world.
How do you personally manage the boundaries between dreams, reality and meta?
I’m not sure what Meta means in this instance. As far as the distinction between dreams and reality goes, as a psychoanalyst, I want to say that dreams are very much reality. We would love to believe that dreams are magical, and come from another place, but in a banal sense, dreams are just another way of thinking about our day. We think all day about our waking life, and we call that ‘reality’ when it’s also made up mostly of thoughts and beliefs that bear only a tenuous relationship to the world. Then we dream all night about the exact same stuff, but because there’s less censorship there tends to be more we’d like to disown.
I’d say AI has the function of the dream; that is, it is a way we can disown responsibility for certain things because the algorithm seems to show us a ‘scientific’ answer to certain conflicts. But the problem with AI algorithms is that they can’t create the future, they can only recombine the past, so an AI is only ever optimising for the situation it learned within.
The best example is the 2007/8 financial crash. It happened because banks were using algorithms to make investment decisions, but the algorithms were trained in a time of great plenty and had no way of dealing with crises. The banks wanted that, because they wanted to imagine the good times would keep going, and for a brief period the algorithms helped sustain that illusion, but that can only last for so long. Then the algorithms fail, and as happened back then, we will blame the algorithms rather than accepting that the algorithms were doing our dirty work for us. Political and social conflicts can’t be solved by automatic processes.
So, the answer to managing the boundaries between dreams and reality and meta (which I’ll take to mean the metaverse or something like that) is, we all already know where the boundaries are and we all already know what personal problems we need to solve and if we try to cede control to the algorithms, it’s not going to go well. I’m from Perth so when I go swimming at the beach, I sometimes imagine I’ll be eaten by a shark, it’s very unlikely to happen, and I could say I am probably focussing on that fear because it’s better than thinking about how much damage humans are doing to the ocean every day and how much harm we’ve caused to the environment that sharks live in. But then there is always also the chance I could actually be eaten.
PHRASER was exhibited at FUTURES Gallery as part of PHOTO22 and was set to run autonomously within the gallery. If you are interested to find out more about Sam’s work, you can head over to his website, https://samlieblich.com to keep up to date with his latest projects.
Written by Beatrice Madamba