Buckhouse speaking at the Sequoia AI Ascent Conference

Augmented Imagination

James Buckhouse

--

Gone are the toll-bridge collectors on the Golden Gate Bridge. Gone are the checkout clerks from your favorite grocery store. But what about us artists? Writers? And thinkers of any stripe? Are we about to get gone too?

AI will not mean the death of artists, intellectuals, or anyone else. Instead, it will mean our rebirth, but only if we make it so. Here’s how: we must stop thinking of AI as Artificial Intelligence, and instead think of it as Augmented Imagination.

If we augment our imaginations with AI — even if our imaginations were mighty powerful to begin with — we’ll create what culture needs but has not yet seen. AI will give artists and designers more power and ability than ever before.

As we augment our imaginations with AI — artists, writers, designers, and filmmakers (both traditional and new) will have new ways to create their best work and remix across the plane of the adjacent possible to show us what we need now (but have not yet seen).

I came up making movies for film franchises that went on to become household names: Shrek, Madagascar, Matrix, and more. I’m trained as a traditional painter and have had gallery and museum shows around the globe. I run a design lab. I’ve designed products on your phone right now. I’ve written jokes you’ve heard. I’m a working designer, writer and artist. Don’t take my word for it… here’s what ChatGTP-4 said in response to Who is Artist, James Buckhouse… you so know it must be true…

I still have one foot in Hollywood, working with writers and directors as they think through story arcs and character development to try to create stories that move us, that transform us. I’m writing as someone who cares deeply about the creation of new art and who hopes that technology will enrich, not oppress humanity, and as someone who cares deeply about the people behind the art and stories that we all love.

And in this moment, I’m exceedingly hopeful, excited, and optimistic.

Why now?

This morning — today, this very day! — is the greatest time to be alive in the history of planet Earth. Yes there’s still war, yes there’s still disease — tragedy abounds. Personal pain persists… Governments struggle, people remain oppressed… horrific atrocities exist… and yet…

And yet.

And yet there has never been a moment when we were more perfectly positioned to live up to our calling to embrace the hard problems of existence and to solve them — and solve them well — than today.

Think of the innovation in the past year. The past month. This past week. In the last 48 hours. This morning is the best morning to wake up alive that has ever happened. The one most ripe with opportunity. But why? Why now? Why not some other time in the last 100,000 or million years?

Our primary adaptation

Today is marvelous because as humans have an adaptation that helps us and has brought us to this moment. We are not the strongest. The fiercest. The fastest. Or the most well-armored.

We lack the teeth of the tiger. The talons of the raptor, the hard shell of the tortoise. We aren’t even the strongest primate. We are essentially sacks of water and meat. No one fears our claws or our teeth. And yet we have one thing. One thing that others do not. We have a special ability. And with it, we can escape the terrible fate of a lifetime of linear progress in our learning — that terrible struggle where you must learn mostly on your own and figure out every problem anew.

Humans do it differently, we can both listen and speak — we can both hear about the experiences of others and share our own so that we don’t just learn what we ourselves have discovered, but yet also we can learn what every single living human before us has figured out to bust our curve of knowledge away from linear growth towards the exponential arc of total accumulative understanding — and then… and then… We can make use of what we’ve learned as we attempt to add our own chapter to the great book of life.

The explosion of potential that happened this morning is the greatest there has yet been. And tomorrow, amazingly, it will be even greater.

And yet what we are doing, today and every day, is more than just learning. We both learn from others and we teach what we’ve learned. It turns out our adaptation is not merely learning — many animals do that — it’s the combination of learning and teaching and remixing that makes us special. And as humans we have a marvelous, portable, memorable, extensive process for this rare combination. It’s called a story. Story is framework through which we both learn and teach and remix.

Story is our primary adaptation. Story is our compact, encapsulated method of knowledge distribution and creation. Story is how we bend the arc of humanity’s progress. It is through story that we fork each others repos and accelerate human potential.

Story is the process that has kept us alive and keeps us flourishing. And it is also at the root of AI. We listen and we learn and we share what we know and we remix to expand the plane of the adjacent possible with each new element that is added in to the treasure trove of human knowledge.

AI does this too—and does it at a scale that finally lives up to the promise of making use of all previous human knowledge. The emergent possibilities are potentially staggering.

But back to Generative AI for art & design for a second… and a brief trip into art history… and as with the entire history of art, to really talk about it we also need to talk about money.

Art Begets Art

In the old days, as artists, we would see work that we loved and it would influence how we created our own. A famous example from 1511 came from a moment when Raphael and Michelangelo were both working on the remodeling of the Vatican. Mike was doing the ceiling of the Sistine Chapel and Raf was working on the papal apartments about 950 feet down the hall. During the day they would pass each other in the corridors like rival cool-kid gangs in high school mocking each other — Mike was too scruffy; Raf, too fashion-focused — and sneer at each other’s work. But at night, they secretly admired each other’s output. One night Raphael snuck into the Sistine Chapel and peeked under the curtain and saw Michelangelo’s magnificent figures. Flabbergasted, he sprinted out and did two things: he scrubbed some of his own figures and made them more like Mike’s — muscular and bold — and then he added an homage to Michelangelo to his own masterwork, The School of Athens, as a celebration of the spectacular work of his frenemy. Incorporating influences from another artist meant doing the work to bring their style into yours and facing up to public scrutiny, recognition, and discussion when you did.

Now, 512 years later, we are undergoing our own rebirth of art. The new possibilities open the mind like the first dawn of spring. And yet with it comes some tough questions of ownership, credit, and payment. Generative AI tools are trained on visual information, and every image comes from somewhere, so if we dig deep enough there is a complicated, but real, DNA of influence for each generated work. With this in mind, could it be possible to find a way to pay artists whose work helped train the model? That sounds hard, and fraught, but likely worthy of investigation.

YouTube worked through a similarly tricky, seemingly intractable, set of issues around people uploading copyrighted information to the platform. They found a novel way forward and, as a result, the creator economy thrived (and copyright holders got paid). Here’s a TED talk on how YouTube approached it. Their answer isn’t perfect in all cases, and doesn’t really apply to AI, but it was ingenious for the problem it was solving. Could something unique, yet equally ingenious be developed for Generative AI?

But what about working now?

While the issues of ownership and payment are being worked on and explored in the courts, there are currently two general approaches that people can use to attempt to address the ownership concerns of models: the first is to use narrow, private, or semi-private models that are trained on your own work; the second is to use licensed or public domain art to train a model.

For example, Adobe states that it only used licensed and public domain sources in the training of its Firefly AI. Further, to quote the Fast Company article on the product launch, “Adobe says it’s working on ways to pay creators whose images are leveraged by Firefly, including contributors to Adobe Stock.”

Another option is for Artists to opt out, at least from some of the data sets. I respect that choice very much, but it has also made me wonder, is opting out the best move? For some it might be.

Can there be a way for AI to work that is fair to artists, propels us forward, and cracks open the magic of this technology to further expand the plane of the adjacent possible?

Immortal Influence

Being an artist is both a lifelong pursuit of excellence and a chance to influence the evolution of culture through your practice… and having your work woven into the DNA of culture is about as close as it gets to immortality.

Jean-Luc Godard, one of the leading lights of the French New Wave of filmmakers, had a line in his masterwork, Breathless, that illustrates this idea: in the scene, Patricia (Jean Seberg), is interviewing an author (played by Jean-Pierre Melville) and asks “what is your grand ambition?” and the author replies “to become immortal, and then to die.” When I first heard this line I loved it. It sounded like a willful contradiction — how can you be immortal and also die? But then as the ice melts in your own mind you realize this is a call to create exceptional art. Transcendent art. Art that survives long past your own life. Art that influences culture and humanity for generations. Art is the path to this type of near-immortal influence.

Still from from Godard’s À bout de souffle

In this way, Generative AI could provide a back door to immortality. In this light, being included in the Generative AI model changes from feeling like a rip-off to being an honor, as it’s a way to have your influence reach far beyond your initial work. When your own art becomes a part of the cultural genome, you’ve secured a place beyond your own reach—but only if you can still eat. I have great hopes that eventually a fair and wonderful solution will be found, maybe through one of the current approaches, or maybe through something new.

In the meantime, professional studios are increasingly thinking about using the art and assets that they already own (or have paid artists to create) to train or tune their models. This idea came up in the recent AI Film festival in NY.

Hear a panel (Paul Trillo, Souki Mehdaoui, Cleo Abram and Darren Aronofsky) discussing the use of AI in film.

Recently I tested out this “more private” approach: I trained an AI on my own paintings and then had it create a few more back for me in my own style. The results are both familiar and strange. Some of these feel very much of my own style and touch. Others, well, less so. And yet, what a strange and wonderful mirror to see myself presented back to me in the form of my own vocabulary. The familiar rendered strange. The self-same as other. I loved doing it. I didn’t feel robbed, I felt marvelous.

It also felt powerful.

Scale of ambition

Using Generative AI, artists could move from creating a single work at a time to being able to create a whole universe of books, paintings, movies, videos, games, and more — all based on their ideas and point of view. When I worked on Shrek, we had 400 people working 11–14 hours per day for 1–4 years (depending on which department you were in). This was a mind-numbing amount of work. We got faster by the time Shrek 3 rolled around, but only a little. It still took 18 months even on the fastest project.

Today with Generative AI, we can start to imagine being able to make a whole movie with just a few trusted collaborators (or even alone?) and do so in a matter of days or weeks, not years. Generative AI holds the promise to transform solo artists into studios. One day, maybe, the stories we watch will unfold before us: unfolding in real-time based on the guidance and direction we whisper into the ear of the model as it generates.

Generative AI tools will change how we make movies, TV, games, artwork and more. Just as the camera obscura, the film camera, and digital painting changed image-making... just as 3D animation and digital FX changed how films are created… just as the laptop changed how we write and the synthesizer, sequencer, and sampler changed how we compose… just like all of these, Generative AI will change how we create.

AI will change art, but it will not kill art, because art is unkillable. Art is as eternal as hope. Art is not a technique or a tool, but an urgent need for seeking meaning in a moment that is ever-evolving. Art is the now. And as the ever-expanding yet ever-receding NOW, art can’t die, even if machines become the mirrors that reflect our new now in ways that get sometimes uncomfortable.

AI will change us, but it will also teach us what we already know. It will remind us what the source of art actually is. It will cause us to swan dive into the abyss of our own essential insights about what matters. The uncomfortable truth is that art is and always has been about the idea — the insight — the emphasis — the λόγος — and not merely the execution of that idea, even as (paradoxically) we only experience the art through its execution. The catch is (and has always been) that we need both. There was even a term for this in the Renaissance. It was called Ars and Invenio. Ars was the craft. Invenio the innovation. You need both a great idea and great execution. Weak, trite, or poor execution will render a great idea inert. And a bad idea with good execution is still bad. We’ve all seen a well-made bad movie. You have to have both. Always. Even the exceptions to this rule are secretly instantiations of it.

Where will AI take us?

A decade ago, it would have been ridiculous to imagine that designers would collaborate on the same designs at the same moments with no-rules-everyone-in-the-same-document-at-the-same-time collaboration. But Figma changed all that, and now we can’t imagine designing any other way. Collaboration just makes all of us much better — because with it we can escape the linear knowledge attainment of a single human’s lifetime of skill acquisition. We instead grow and learn at the speed of everyone’s learning, not just our own. The change waiting for us with Generative AI is even greater.

Generative AI has a shot to transform everyone into an image-maker, and every professional artist into a studio. It has a chance for all of us to learn at the speed of everyone’s insights to get to the truth of the human condition in ways that have not yet been seen or experienced. It gives us a shot to chase our curiosity all the way to revelation and grasp at greatness on the way to cultural semi-immortality. It is a chance to weave ourselves into the global, cultural DNA while benefiting from and contributing to the strengths of everyone else’s efforts. It’s a chance to transform the world again through a new type of ars and invenio. It’s a chance to make a difference.

--

--

James Buckhouse

Design Partner at Sequoia, Founder of Sequoia Design Lab. Past: Twitter, Dreamworks. Guest lecturer at Stanford GSB/d.school & Harvard GSD jamesbuckhouse.com