Do Bots Know Beauty?

Alan Turing proposed that we judge the sentience of AI programs based solely on their ability to mimic convincingly human responses to a battery of questions posed by a trained examiner.

Recent advances in AI technology suggest that we will soon have, if we don’t already, machines that can pass Turing’s Test. (On the other hand, I have doubts about some of my carbon-based neighbors…but that is another subject for a different day.)

Are we willing to accept the results of Turing’s Test as definitive? Or will we ‘move the goal posts’? Now that you’ve learned your ABC’s, Sally, please recite them again, but this time in reverse order and, oh yeah, in ancient Greek…if you don’t mind.  (We’re not going to let some snot-nosed rugrat show off in front of us! After all, we can all recite the Greek alphabet backwards, can’t we?) So come on, Sally; get crackin’: Omega, Psi…”

Turing devised his ‘test’ in the context of the materialism, pragmatism, and behaviorism of his time. It is understandable that we might now want to review Turning’s proposal in the context of a more contemporary, more expansive, ideology.   

It is interesting how an issue that forms the cornerstone of one school of philosophy can be discounted as a mere distraction by a different school. At Aletheia Today we are admittedly ‘hung-up’ on the question of values – objective, transcendent values. If such values exist, then an ontologically flat, wholly immanent universe (a la Nietzsche) is simply not possible. 

On the other hand, if such values do not exist, then ethics and aesthetics are just matters of taste…if that. Without objective, transcendent values, free will becomes caprice. In fact, it is hard to see how any events can ever occur in a universe bereft of motivation, incentives, objectives, and purpose. But not everyone agrees with this analysis.

So, let’s get granular! I recently encountered a work of ‘art’ (above) that is almost certainly computer generated. I know nothing about the provenance of this piece, but judging by the style and content, I can only assume that an AI Bot was given specific input re themes, images, etc. and then allowed to ‘create’. Nothing is ‘wrong’ with the result…except that it is not beautiful.

The Bot followed instructions, and produced this ‘Rube Goldberg’, all according to spec. The piece contains some interesting local details and flourishes – geometric shapes, meditations on gravity, modulation of colors - but none of it hangs together. The piece has competent parts, but there’s no sense of a ‘whole’, perhaps because that would require the Bot to step back and consider its work as a whole…which may be something that this Bot cannot do.

The Golden Mean notwithstanding, do we think it is possible to reduce Beauty and Truth, “all ye know of earth and all ye need to know” (Keats) to an algorithm? Actually, we discovered part of that answer while Alan Turing was still in school. In 1931, Kurt Godel proved that mathematics could not be reduced to a system of algorithms: ‘There are more (true) things in heaven and earth…’ than can be proven in any logical system.

According to John Keats (above), “Beauty is truth and truth beauty.” I agree with Keats, but I would modify his dictum with a nod to Alfred North Whitehead. Beauty is the broader, more inclusive category; Truth and Justice are specific (albeit peculiar) manifestations of ‘Beauty’ (harmony, consistency, etc.).

Do Bots know beauty? Can they create something unique that is also ‘beautiful’? Of course, they can! They’re doing it today…and have been for a couple of decades now. Contemporary artists use the microprocessor as we would a paint brush…and they get similar results. But the ‘beauty part’ comes entirely from the artist. The artist creates, the Bot executes the artist’s instructions. The relevant question remains, “Can a Bot create Beauty on its own?” 

An artist can step back from a canvass and say, “Eureka, it’s finished!” or “Nope, not yet.” Can a Bot, on its own, make a similar judgment? 

They say the soul of every artistic creation, in any medium, is editing. Can Bots edit? Can they look at a perfectly coherent finished product and say, “Something’s not right here” or “something’s missing” or “it’s too busy” or “it doesn’t read well”? 

One of the hardest parts of being an artist is knowing when something is done, finished, complete. There is always a temptation to change a word, add a brush stroke, alter a chord; how do you know when to stop? How does a Bot? Yet knowing when to stop is a critical component of the creative process. 

I don’t propose to give a definitive answer here. I’m guessing Stanley Kubrick (2001: A Space Odyssey) would answer these questions differently from Roger Penrose (The Emperor’s New Clothes). I do propose, however, that we make this the test, not Turing’s, that determines whether a Bot is conscious.

Have you ever stood on uneven pavement where one foot was slightly higher than the other? It’s unsettling to say the least. That, in my view, is the origin of consciousness. Our mental apparatus compares immanent ‘reality’ with the transcendent values (Beauty, Truth, and Justice) that govern the Universe; the perceived imbalance gives rise to conscious experience.

No values, no consciousness! Therefore, demonstrating that a Bot, on its own, can identify what’s beautiful, true, or just, absent specific operator instructions, would satisfy me that the Bot is conscious, i.e., that it can recognize and apply Value without operator prompts. 

My test is easy to devise but much harder to conduct. Bots are master mimics. When we think of mimicry, we think of monkeys and parrots. (However, we now suspect that their behavior is less mime and more consciously chosen behavior.) 

Bots, on the other hand, are built to mimic. They can ingest and regurgitate a 12-year curriculum in seconds. Then they can adjust that knowledge base as new facts are added and even as teaching methods modernize. 

Take, for example, the picture shown above. I could tell my Bot in my most avuncular tone, “This is a good start, but it’s not quite what I had in mind. Try blending the colors a little more, add some more interesting background detail, work the foreground figure into the tableau.”

My Bot and I could repeat this process ad nauseam, eventually getting to a point where I might say, “Ok, we’re done here. Good job!” Is this ‘creation’? Can it be ‘beautiful’? Yes and yes! But is the Bot the creator…or am I? 

Beauty, Truth, and Justice are transcendent values. They are manifestations of eternity in spacetime. Stepping outside of Nietzsche’s ‘ontologically flat world’ to view that world as a whole, accessing transcendent values and applying those values to the situation at hand – that’s consciousness.

I believe that I have the capacity to do just that, and I take it as a working hypothesis that I share that capacity with most other human beings and with at least some other terrestrial life forms. But how about AI Bots? Do they share this capacity? Can they?

David Cowles is the founder and editor-in-chief of Aletheia Today Magazine. He lives with his family in Massachusetts where he studies and writes about philosophy, science, theology, and scripture. He can be reached at david@aletheiatoday.com.

Return to our AI Issue Table of Contents