Can Machine Creativity Truly Exist?
Five curators, artists and writers discuss the impact of AI models like ChatGPT on artistic production
Five curators, artists and writers discuss the impact of AI models like ChatGPT on artistic production
Ben Davis How are you all feeling about the new wave and conversation around generative AI and LLMs [large language models] like ChatGPT? Are you positive, negative, some mix of the two or do you think it’s just a flash in the pan?
Mike Pepi I am of the mindset that, despite the venture capital-fuelled hype, LLMs are essentially just a parlour trick. They will never have true sentience or creativity. And I think it’s very important for anyone working in a creative field and feeling anxious about AI disrupting their output to just temper some of those utopian prognostications that come from the people who put these things out. I think we’re at the very tip of the hype cycle and soon we’re going to see that we prefer to get things from humans.
Noam Segal Maybe we should start by saying that when big LLMs only operate on their own generated data, the system collapses, it leads to a system breakdown. If we think about the writers and actors currently striking in Hollywood, for instance, they cannot fully be replaced by AI because their work cannot be reduced to data performativity and these models cannot yet generate causality or react to context. I like to think of it as a prosthetic tool, like Yuk Hui suggests. AI as a tool can be very helpful in generating versions of existing things; when thinking about virus mutations and vaccines, it’s a gamechanger. We really don’t need to be anxious about it: AI can feed back and add different layers and options, but it cannot replace humans.
Stephanie Dinkins I think it’s hard to put AI into a negative or a positive camp because it exists in our world in a way that’s quite impactful. But I don’t think these LLMs, for instance, will be around that long. Something else will come along and supplant them and we will have to learn what that does and navigate it.
Christopher Kulendran Thomas I tend to think of the AI and ML [machine learning] tools I use as exactly that: tools. There are things I couldn’t make without those tools. For instance, parts of my video installation The Finesse  are algorithmically auto-edited with footage scraped live from social media, so that it’s different every time it loops, and it’s narrated by a generated avatar of a well-known celebrity who has been trained on more texts than any human could ever read, so I never know what they’re going to say. In a few different ways, that work has no fixed perspective. The viewer has a different experience each time they walk into the space and that wouldn’t be possible in the same way without these tools.
I think though that this conversation is one that could only happen today because it will very quickly become something we won’t need to talk about – at least any more than we talk about paintbrushes or camera lenses. I think it’ll quite quickly become more interesting to talk about what you do with the tools than the tools themselves, because the AI and ML tools we’re talking about now are becoming ubiquitous. With that ubiquity, though, will come a shift in perception that I think will be quite profound – and that will be interesting to talk about.
I am of the mindset that, despite the venture capital-fuelled hype, LLMs are essentially just a parlour trick
BD I think there is a very big difference between the use of AI in art and the implications of AI in society more broadly. In my opinion, art is a very eccentric case. In general, I feel extremely negatively about AI because the economic structures that I am bound up with as a writer are defined by an acceleration of output and a degradation of labour conditions that I can only see AI worsening. I don’t see it freeing up space for creativity at all within the world of writing. Not that it couldn’t.
MP I agree. I think the problem here is that we live in a capitalist society and we’re playing by those rules. There is a theoretical communist AI. There is a theoretical humanist AI. But in the industrial creative space where there is a one-to-one exchange for your labour – fields such as graphic design, for instance – things are going to get bad, if only because the right institutions aren’t there to stop it.
NS I agree that we should look at AI in the same way we look at a paintbrush or a camera. At the same time, however, it does require a new language because the material of AI are datasets that are generated and extracted from all of our online procedures and operations. In that sense, capitalism is an intrinsic part of it. It’s conceivable that the courts will determine whether AIs truly mirror societal agency through their datasets and coding, or primarily represent the financial interests of stakeholders, thereby treating corporate AIs like any other commercial enterprise.
CKT I guess there is a utopian fantasy of a post-capitalist economy that comes about through artificial intelligence, nuclear fusion, genetic engineering and so on. And I think it’s worth noting that these technologies are not only being developed by liberal capitalist democracies. It’s probably beyond the scope of today’s conversation whether you consider the Chinese Communist Party a capitalist government or not. But the political system of the West is not the only context in which AI tools are being developed. So there’s obviously a geopolitical dimension to it, too.
SD I would second that. I think these technologies require us to examine our systems deeply. Of course, this is long-term work.
BD Do you think AI might ever provide solutions to inequality or social unrest?
AI can feed back and add different layers and options, but it cannot replace humans
SD I hope so, but I don’t know that AI is going to help us. We want to offload our problems onto this technology, but it always comes back to the humans behind it and how we start those negotiations to even get the AI to a place where it might help us.
NS I think this geopolitical dimension is important to address. For example, I’m now working with LG and their AI really is designed differently and answers questions differently from ChatGPT; in the same way, Stephanie’s AI work, Not The Only One , also offers another set reactions and conversing with her leads to a more intimate dialogue. We tend to bundle them together, but there are various kinds of machine intelligence because altogether it reflects the designing code and its collective feedback.
CKT Our current political and economic system is based on the idea that we’re all individually and ontologically distinct from everything that’s not us. In that sense, you could see that idea of individualism as one of the most profoundly important psyops [psychological operations] the West has ever devised. But it’s a fiction that is unravelling both technologically and also geopolitically, in the shift and in the balance of power from the West to the East. I think ML tools give us one way into thinking about who we are in the world beyond this fiction of individualism.
MP I’m a humanist, but I agree with Christopher’s point. I believe the task of the critic is to potentially bear witness to that unfolding. However, I think we’re getting into more of a political, philosophical debate around what we want to work for. I don’t quite know if that’s art’s role, really – nor even the ambit of this conversation.
BD I think it’s all connected. There is a way in which these things are just tools with a human behind them. And, like Noam was saying, different human decisions have gone into creating them. On the other hand, to connect with what Christopher just said, I do think that the agency of the individual is very limited by our societal and economic structures. There’s a difference between a regular tool used by a person and a tool like AI whose function is determined by the structures within it. It’s like the illusion we have that CEOs make individual choices that determine how a company functions and treats its workers. But, ultimately, they’re constrained by the logic of how they’re funded by the stock market and capital. The CEO could be a good person, but the incentives are aligned for them to do a certain thing in a certain way. I think that will affect these models, too, regardless of the individuals using them.
NS I guess, to a large extent, what is amazing in AI is the creation of everything that is calculable. And, for that, I really believe it’s going to find a cure for cancer because it can do more calculations than any other tool we have. Is this creativity? Depends who you ask. However, when we think about the results, the generative outcomes of AI, I find that it’s valuable to differentiate between the things that are calculable and the things that are not.
It’s very hard to think that human phenomena are calculable. And we can think about this in relation to legal proceedings. In China, a lot of the judicial system is already automated into AI systems – something that might be implemented here in the US. The thing to focus on is what these tools are best for and where they can help us while remaining tools rather than recursive beings that have equal agency to us.
SD I try to remain a humanist as well, Mike, but I feel myself sliding into the middle in terms of where I sit. And I’m certainly not a human supremacist. A lot of our current thinking is based on the idea that humans are still cream of the crop. I don’t know how long that belief will last.
MP I agree: I think it’s really hard to use binaries here. Clearly, something in the middle is possible.
BD Something you said originally that I maybe disagree with, Mike, is that these technologies aren’t creative. I think they’re very transparently, transcendently creative – and more naturally creative than 99 percent of people, just with what they can do now. But part of the human experience is being fragile and limited. The fact that there are only so many paintings by Vincent van Gogh is a feature, not a flaw. If we use your work as an example, Christopher, what does it mean for an art audience not to have a common experience of an artwork? People have shared experiences of interactive media all the time: video games are a good example. But what does the collective nature of art and the limitlessly customizable nature of some of these creative objects suggest about how you review or receive them?
MP I think the essential point of this discourse is: do we accept machine creativity as having its own philosophical ideas and biases? It’s not as if humans have already left the scene and are therefore absolved from investigating this. It’s fair to say that humans alone currently rely on these institutions and collective ideas, Western or otherwise, but then the question becomes: are we giving a free pass to machine creativity just because it’s not in this enlightenment humanist paradigm? We need some philosophical framework that keeps motivating this concept.
NS I agree. When we say creativity, doesn’t that necessarily imply intentionality? Can we even use the term ‘machine creativity’? The notion that creativity can be as simple as connecting two ideas that you have already heard and spitting them out feels a bit glib. But that is exactly what AI is doing. The concept of metaphoric thinking might be useful here, because this is also what we do when we create metaphors, for which, in turn, we create new meanings. And those new meanings come through experience, through living.
This is where machine creativity differs from human creativity. To me, it falls to questions of intention and causality. AI can combine two things and, perhaps, after generating 100,000 results, one of them could qualify as a metaphor, but that’s due to probability and chance.
With that being said, we’re getting closer to the day when each of us will have our own unique artworks created by a proprietary AI, and the very loose framework or definition that we have for an artwork will be lost altogether. Unfortunately, at that point, this idea of a collective experience will no longer be part of the discussion. It will just be a question of whether something custom-made for an individual can bear the same articulative conditions as an artwork.
SD There’s something beautiful in what you’re saying, Noam, which is that we’re regressing to the point where individuals will have agency in their own creativity versus having to go through a set of gatekeepers who designate what good art is. That is so exciting.
You can already see how that has played out in television, which used to be a collective experience – shows were watched in a communal space at the same time – and also in religion. Churches, which were meant to communicate to the masses and share a common set of values, are used less and less. So, how do we get back to something that gives us that? And what are the rules?
BD To return to Noam’s question: does creativity imply intention? I don’t think it does. I would argue that evolution is a creative process, but there’s no intention to it. In terms of the reception of creativity, however, the perception of some social meaning is important.
I’m not that excited about infinite customization. I think it solves the wrong problem culturally. Every sign that I see points to the fact that there’s too much culture and not enough community around it: we have a tremendous overproduction of stuff for a very small audience. There’s also all this data about the long-term decline in social organizations of all kinds: churches, unions, even friendships. To a certain extent, parasocial relationships have replaced real ones. And that’s part of what is being sold in a lot of these technologies.
CKT But, Ben, I’m not sure that that’s the only conclusion you can draw from the idea of infinite customization, because it may just be a question of scale on the level of the organizational form itself. Broadcast media, for instance, cohered nation states and large communities that spoke the same language, transacted in the same currency and broadly shared the same culture.
BD You mean Benedict Anderson’s notion of Imagined Communities ?
CKT Exactly. I think the fragmenting of media will certainly undermine that organizational form, but I don’t think it automatically means that other kinds of communities can’t cohere through that fragmentation. The Web 2.0 internet era has had these very individualistic media forms that, in turn, have produced the influencer as the archetypal business model. But it’s also interesting how we’re beginning to see online subcultures almost collectively hallucinate a scene into existence. I hate to use the word ‘community’ because it’s such a bullshit, oversignified word used by every lame e-commerce brand that doesn’t actually have a community, but the idea of some synonym for a community that has been collectively hallucinated into existence through some kind of shared, algorithmically mediated spirituality – an egregore if you will. I find that pretty interesting.
You could see that idea of individualism as one of the most profoundly important psyops the West has ever devised
SD I think a lot about the various myths that we tell each other – and ourselves – and I truly believe that the use of AI will help eliminate these. I speak to a lot of folks who criticize the technology but they’ve hardly ever used it or seen what it can do – how it could add to the discourse around racial equity, for instance. Then, when they do, and they’re given a little help to achieve something different, they’re surprised at how well it can be used in their line of work, or how well they can manipulate the tool to do something that they want it to do. There are tons of folks out there who are calling out the problems and biases. I always ask people to use the technology and really dive in and figure out how we might skew it.
CKT If conversational interfaces are what comes after the web browser, then the UX challenge is going to be a kind of race to intimacy. And I think the potential for propaganda and thought control there can run so much deeper than any media system we’ve ever seen. Just the potential for censorship on a whole other scale seems so inevitable.
SD That’s where I go, too, Christopher, and that’s why we have to be proactive. Our only alternative is to imagine the unimaginable.
CKT So maybe the unimaginable thing, the societal transformation we’re really facing, is about letting go of the dual ideas of ‘free will’ and ‘reality’. Maybe this was a fiction from an empire anyway.
SD That’s interesting. Yes, letting go of it and then figuring out how we use some of that power and mind control ourselves. We’ve all seen it being enacted, especially here in the US. In some instances, it’s being done by powerful entities. In others, it’s being done by people who are amassing power. If a brown man can lead a white supremacist group like the Proud Boys, then what’s available to us?
CKT Well I guess that’s the crazy thing about being alive right now – it does seem like anything could happen.
This article first appeared in frieze issue 239 with the headline ‘The Face of the Deep’