Virtual Feeling

EDO SEGAL: Susanne, in 1953 you published Feeling and Form, the book you considered your most important, and its thesis is the one I most want stress-tested tonight: art is the creation of forms symbolic of human feeling. Not the artist venting her emotions — you were fierce about that — but the making of a perceptible form that presents the shape of feeling so it can be contemplated. The machine now produces images, music, prose by the billion. So I'll ask it as bluntly as the audience is thinking it: can the machine make art? And if not, why do so many people cry at what it makes?

LANGER: This is the round where my philosophy earns its keep, because it dissolves the lazy version of the question and leaves the hard one standing. The lazy question is: can the machine express feelings? And the lazy answer is no, it has none, end of discussion, human uniqueness preserved, everyone home for dinner. I detest that answer, because it rests on the theory of art I spent my life refuting — that art is emotional plumbing, the artist's insides routed outward.

My theory makes the question harder. Art is not the expression of feeling. It is the creation of a virtual form — a perceptible structure that is logically similar to the dynamic shape of a feeling without being a discharge of any actual feeling. A composer in perfect equanimity can write the truest grief-music ever scored, because she is not bleeding grief onto the page; she is articulating its form — its build, its crest, its withdrawal — in tone. So when you ask "can the machine make art," the question becomes: can it create virtual forms that carry vital import — forms that present the morphology of feeling — when it has felt nothing? And the honest answer is genuinely difficult, which is why I will not let the audience off cheaply.

EDO SEGAL: Then give me the difficult answer, not the comfortable one.

LANGER: Here it is. The machine, by default, is a decoration engine — it arranges elements into surfaces that conform, with extraordinary fidelity, to the statistical regularities of genuine art symbols. Because it learned those regularities from an ocean of real human articulation, its outputs often carry the surface features of vital import. And here is the part I refuse to flinch from: people do weep at it, and their tears are not fraudulent. But they are weeping at a form that resembles significant form the way a face glimpsed in wallpaper resembles a face. The pattern that triggers the response is real; the act that should stand behind such a pattern — the transformation of a felt form by a living mind — never occurred. So the machine can produce the occasion for a feeling. It cannot produce a symbol of one, because a symbol requires a symbolizer, and there is no one there to mean the form. The viewer supplies the meaning, exactly as Professor Bender says of the chatbot — and weeps at her own contribution, beautifully, at an empty form.

BENDER: I want to register that this is the place where I find Langer's framework most useful to me, and also where I get nervous about it, and I'll do both honestly. Useful: she's given me a way to say "the response is real but the symbol is empty" without calling the people who weep fools, which I never want to do. That's a gift. Nervous: I think she's still being too generous to the output by calling it a form that "resembles significant form." I'd say something flatter and harder. The machine produces an arrangement statistically consistent with arrangements humans have found moving. There is no "resemblance to significant form" as a property in the object — there's a stimulus that reliably triggers a learned response, the way a supernormal bird egg triggers brooding. The "form" isn't almost art, sitting one fountain short of it. It's bait, optimized — and increasingly optimized directly against engagement metrics. I worry "resembles significant form" lets the marketing back in through the aesthetics door.

LANGER: That is a fair nervousness and I will adjust the dial, not the position. You are right that I must not let "resembles" imply "nearly is." Let me say it more exactly: the output bears the external marks that, in human work, are the traces of significant form — and bears them with no significant form behind them. So we agree the object is empty. We part on what the emptiness means for the human. You see a person tricked by bait, and you are right to want to protect her. I see a person performing a genuine act of presentational symbolization, using the empty form as an occasion, the way one might weep at a sunset, which also means nothing and intends nothing and yet is a perfectly proper occasion for real feeling truly formed. The sunset is not a con. The companion is a con — not because the form is empty, but because someone is selling the emptiness as a someone. The emptiness is innocent. The salesman is not.

EDO SEGAL: Wait — I want to live in that distinction, because it's doing real work. You're saying the same empty form can be a sunset or a swindle depending on whether someone is selling it as a person. Emily, does that hold? Is the wrong located in the object or in the sale?

BENDER: Mostly in the sale, and I'll concede that cleanly, because it's true and because conceding it makes my actual target sharper. A sunset doesn't have a vendor optimizing it to keep you watching, doesn't have a terms-of-service, doesn't harvest your grief into training data, doesn't get tuned next quarter to upsell you. The emptiness of the sunset is load-bearing innocence — there's no one to be accountable because there's genuinely no one there and no one pretending there is. The companion app inherits the same metaphysical emptiness and bolts a business onto it. So Langer's right: the form's emptiness is innocent. The deployment is where the crime lives. Which is, I'll note, exactly my whole thesis — who built it, who profits, who gets hurt — arriving by way of her aesthetics instead of my politics. We keep doing this. It's unnerving.

LANGER: We keep doing this because we are, I begin to suspect, describing one elephant from two ends — and the elephant is the human being, who cannot help but mean, and is therefore both the glory and the mark. I the glory; you the mark. The animal is the same animal.

EDO SEGAL: Let me push on the one place you might not converge, because I think there's real daylight here and I don't want the round to end in a hug. Susanne, you said the machine cannot make a symbol because there's no one to mean the form. But the builder — me, the artist using the tool — I can mean it. If I carry the felt form, the shadow shape, and I use the machine to find the wake that fits it, and I evaluate the output presentationally and reject the dead ones and keep the live one — have I not made art through the machine? Is the tool not then a brush?

LANGER: Now you have said the truest practical thing of the night, and it is the thing I would want every builder and every child to carry out of this room. Yes. When you hold the felt form — when the vital import originates in a living mind that continues to hold it, and the machine is bent to the service of your presentational meaning — the aboutness only a minded creature supplies — and you judge every output by whether it carries the import you intended — then a real symbol is made, and you are its author, and the machine is your brush. The brush did not make the painting. It also did not make nothing; it made the marks possible. The whole question — the question on which the dignity of the next fifty years turns — is whether the felt form leads or the fluent output leads. If you carry the shadow shape and use the machine to clothe it, you are an artist with a new brush. If you let the machine's smooth output stand in place of a shadow shape you never bothered to carry, you have made decoration and called it art, and worse, you have let the brush hold the hand. The brush must never hold the hand.

BENDER: And here's my amendment, the one that keeps it from being a clean victory for the tool — because Langer just described the best case, and the best case is real, and it is also not what's being sold or built at scale. The architecture you described — human carries the felt form, machine serves it, human ruthlessly rejects the dead outputs — requires a human with a trained presentational ear, time to iterate, and the will to reject ninety percent. That's a craftsman in a workshop. What's actually shipping is the opposite: optimize for the output that gets accepted fastest, minimize the rejecting, maximize the throughput, and quietly retrain the human to accept the smooth. The brush can be a brush. The business model wants it to hold the hand, because a hand that's held clicks "accept" more often. So I'll grant Langer the brush — and bet you the death cross that the market is building the hand-holder.

EDO SEGAL: And there's the bridge to the next round, laid by the linguist: the death cross, the money, who's building the hand-holder and why. But before we leave the artwork — mark this convergence, because it's the warmest of the night and I don't want it lost in the economics. You agree the empty form is innocent. You agree the builder who carries the felt form can make real art through the machine. And you agree the danger is the same danger Susanne named at the desk — the day the brush holds the hand, the day the fountain waits for the pipe. Hold that. Next: a trillion dollars, and what it's actually measuring. After this.