What do we need art for when machines can think?

YuanSupporting things that leave no physical object behind has been part of my collecting practice from the beginning. Since I started collecting in 2011, that has meant supporting artists’ projects, museum exhibitions, the Italian Pavilion at the 2017 Venice Biennale, as well as making donations to the Metropolitan Museum of Art. So if you say that I have become more interested in “making something possible” than in possessing the final result, that does touch something fundamental, but it did not come from a single turning point.

It is more connected to how I understand life. Exploring an unknown possibility is simply more fascinating to me than possessing a predetermined outcome. Collecting has therefore never been an endpoint of ownership. For me, it is a dynamic process: a continuing cognitive journey through artworks, markets, historical trajectories, changing contexts, and conversations with other people. It is a way of learning.

YuanTraditional collecting tends to remain closer to the individual. Patronage is more explicitly a social act. As an individual, one is only a small part of a much larger ecology; patronage becomes a way of weaving oneself back into society.

But it immediately raises a question of intention: What is the purpose of your support? The absence of an instrumental purpose can itself be a very conscious position. Support, then, becomes an inquiry into what might emerge when you refuse to guide the hand of the maker. In that case, perhaps the rarest quality in a patron is restraint — the willingness to surrender control to the creators and to variables you cannot predict.

But if you do have a specific purpose, the responsibility changes. You have to think seriously about how to realize it, mobilizing whatever knowledge, resources, and relationships you have to build the conditions for it.

YuanI borrowed the metaphor of the “observer effect” from quantum physics because the difficulty in interdisciplinary collaboration runs much deeper than a simple mismatch between different ways of thinking. At a more fundamental level, when people remain trapped inside their own cognitive frameworks, they can only interpret an unfamiliar field through their existing frame of reference. On the surface, two people may appear to share the same conversation, but each is simply reducing a far more complex reality back into a familiar version of what they already know.

AI has drastically lowered the technical barrier to exploring another discipline; access to information is no longer the bottleneck. The real challenge is whether an individual is genuinely willing to interrogate their own premises and broaden the lens through which they view the world.

A good institution does not rush its participants into consensus, nor can it pretend to be neutral. Simply being “interesting” is not enough. It must anchor the core values that guide the inquiry, curate the friction between ideas, while exercising restraint — knowing when to step back, wait patiently, and observe whether this collision reveals an entirely new possibility.

YuanNot everything needs to be translated or aligned immediately. Of course, the context matters. If you are producing an engineering or commercial project within a fixed time frame, you have to reduce ambiguity. You need efficient translation and a common language. But when you are exploring an unknown frontier in art, culture, or science, the opposite may be true.

Many meaningful collaborations do not begin with consensus. They move through a cycle: contradiction produces reflection; reflection leads to adjustment; adjustment generates a new contradiction; and then another period of reflection begins.

From that perspective, misunderstanding and cognitive difference are not simply obstacles. They can become sources of tension and creative energy. Sometimes allowing different ways of seeing to remain different, and even learning to enjoy that untranslated friction, is exactly what allows something to emerge that neither side could have imagined beforehand.

YuanYes, absolutely. When a world becomes entirely frictionless, it also becomes devoid of depth. Friction is not an engineering flaw to be solved; it is an essential cultural condition that keeps our thinking alive.

YuanI think it activated a form of nonlinear thinking in me. Much of our conventional professional thinking is highly dependent on logic and causality. When I stood in front of that large Joan Mitchell painting, I wasn’t consciously trying to extract any particular insight from it. But something in that visual experience seemed to reorganize the topology of my thinking at a subconscious level.

I wasn’t aware of it at the time, but a week later, that shift suddenly surfaced in a completely different context — it helped me find a solution to a systems architecture problem that had been troubling my team for months. The experience also made me more aware of the limits of what language can reach.

We tend to equate knowledge with what can be clearly articulated in words. Yet certain experiences cannot be compressed into symbols, words, or explicit information, even though they can still be passed from one person to another. A Zen Buddhist idea touches on something similar: language is the finger pointing at the moon, not the moon itself. Beneath information and logic, there may be another current of perception and experience. It may not be “transmitted” straightforwardly, but under the right conditions, it can resonate from one person to another.

YuanFirst, we need to ask what we mean by “data”. What we call data today is largely a representation of reality that has been made explicit, discrete, and symbolic so that it can be processed by contemporary computational systems. Human experience is much wider than that. Sensory perception, embodied memory, and intuitions shaped over extraordinarily long evolutionary timescales contain far more than what we conventionally describe as data.

So yes, some very important forms of human knowledge have never become data. But if we take your idea to its logical extreme, the question becomes even more compelling.

Imagine that future biological sensors and brain-computer interfaces become capable of translating ever-deeper layers of sensory experience, subconscious neurological activity, or biological information into data. Machines might eventually become extraordinarily sophisticated simulators or amplifiers of human perception. They could perhaps understand the triggers of human sensation more precisely than we do, or transform what we currently call “intuition” into calculable parameters. We cannot dismiss the possibility that this could happen. The deeper, open question for me is what purpose — at an ontological level — taking that step would actually serve, for either humans or machines.

YuanWhen I speak about a break in rhythm, I mean a widening gap between the speed of technological change and the time required for social and psychological adaptation. Earlier technological revolutions unfolded through physical infrastructure — railways, factories, power grids — which naturally allowed society time to adapt.

AI operates at bit speed, scaling globally through invisible network architectures. Personally, I embrace AI; it expands my capabilities. But we cannot mistake the optimism of a tech bubble for the reality of the broader social fabric.

Consider the growing pushback across Europe and the US against the massive power and water consumption of hyperscale data centers. On the surface, these are conflicts over resource allocation. In reality, they are a visceral societal immune response — a reaction against an unseen, abstract force reshaping everyday life faster than society can process.

The time society needs to digest ethical implications, negotiate public governance, and allow individuals to find an inner anchor cannot be compressed to computational speed. If technology accelerates while human reflection is bypassed, that break in rhythm will rupture the physical world. Some things must deliberately remain slow.

YuanTrust. We are entering an environment in which text, images, and voices can all be generated and convincingly simulated. Production becomes more efficient, but authenticity becomes harder to establish. Real trust depends on face-to-face conversation, on navigating uncertainty together, and on the accumulation of vulnerability and honesty. And all of that takes time.

YuanYes, because I am not sure that machines should become the reference point through which we judge whether human activity remains necessary. Humans create, reflect, and adapt out of drives intrinsic to our own existence. We have never done these things because we were the most efficient tool in the universe.

When we use machine capability as the measure of whether human activity is still justified, we place two very different forms of intelligence into an imaginary competition. And so we grow anxious whenever machines outperform us in one domain or another. Why should that determine our value?

The practical question is different. Will machines replace many forms of human labor? Absolutely. That transition will be painful. Jobs will disappear, and new forms of work and collaboration will emerge. We urgently need policies, institutions, and social mechanisms that can support people through that transformation. But that is fundamentally different from asking whether human life still has purpose.

YuanI wouldn’t say it is the wrong question. It expresses a very real anxiety. But if we look beneath that anxiety, we realize it stems from a fundamental misunderstanding of our own origins.

From an evolutionary standpoint, our ancestors did not stride out of the forests onto the dangerous savannah as noble explorers. They were cast out — forced into exile by stronger primates in the competition for forest canopy. Bipedalism, brain expansion, toolmaking, and language were essentially the survivor bias of evolutionary outcasts enduring catastrophic shifts through pure contingency and adaptation. We were never engineered to fulfill an assigned function. So when we ask about human meaning in the age of AI, we must stop projecting that instrumental demand onto ourselves. A machine is optimized toward an explicit loss function. A human life arrives with no such objective. We experience, suffer, love, and leave marks on cave walls or canvases not to demonstrate efficiency to another intelligence, but because finite life has an inherent urge to express and establish itself.

When instrumental labor is automated away, we are left with a bare existential question: In an accidental universe without preassigned purpose, within a brief and irreversible life, how do you choose to anchor your mortal body? What do you want to experience, and what or whom are you willing to take responsibility for?

Measuring ourselves against machines is often just a defense mechanism to avoid the terrifying solitude and freedom of that question.

YuanWhen I say that technology projects a particular culture, I mean that today’s technological development already carries a very powerful set of values: efficiency, scale, speed, capital, growth, competition.

Sometimes we speak as though technology itself has escaped from human control. But perhaps what has become difficult to control is not technology in isolation; it is the dominance of a particular value system around it. When speed, scale, growth, and monetization become powerful enough, they can begin to appear not as choices but as historical inevitabilities. But they are not inevitable; they only represent one set of coordinates. Turning our gaze toward culture allows different forms of perception, different lines of questioning, and different measures of value to remain present and endure. It reminds us that what currently appears to be the only path forward is, in truth, merely one path among many.

Yet we must also remain clear-eyed about the scale and pace at which such influence operates. Cultural change is slow; it requires people willing to work through these questions over the long term. In the more urgent present, humanistic thinking cannot linger solely at the level of critique or aesthetic reflection. It must actively inform policymaking, shape the architecture of platforms and institutions, and embed itself in the practical mechanisms through which society negotiates technological change.

YuanIn some ways, the world you are describing has already begun to arrive. Recently, I saw Jon Rafman’s exhibition at K21 in Düsseldorf, and it deeply affected me. In an exhibition like that, you see the true reach of AI when placed in the hands of an artist with profound imagination and insight.

So I feel no nostalgic need to encounter work made exclusively by human hands. At the moment, I still find that purely algorithmically generated work rarely reaches the complexity, depth, or multidimensional resonance of great art. But I wouldn’t rule out the possibility that this could change.

Art history has heard declarations of the “death of painting” repeatedly for almost two centuries. Generations of critics who pronounced its demise have passed away, while painting remains. It has survived not by refusing change, but by continually redefining itself through each crisis.

From cave walls and mineral pigments to oil paint, photography, neural networks, and code, the medium keeps changing. What persists is the human impulse to perceive the world, express uncertainty, and explore what we do not yet know.

YuanWhen I was young, a painter once told me, “A true painter, standing before a blank canvas, feels a primal urge to make a mark.” Later, I began to recognize the same impulse in people across science, technology, and other disciplines. A common pulse runs beneath them all.

Perhaps, early in life, a teacher or mentor can help ignite that spark. But fundamentally, I think the impulse itself is instinctive. It belongs to the person.

Knowledge can be passed on. Methods and techniques can be taught. But that primal impulse — the vital spark behind true creativity — cannot simply be manufactured. It has to awaken from within.