Ken's Blog

One frustration about doing computer graphics is that you are essentially creating art with math. There is nothing inherently wrong with that. In fact, it’s a totally wonderful way to spend your time.

The problem comes in when you want to talk to people about how you did it. Nearly everybody loves art, but nearly everybody is afraid of math.

It might have something to do with the way math is taught in school. Instead of being taught as something fun and exploratory (which it is), math is often taught as a set of strict rules — if you get something wrong you fail the test.

But that way of teaching really has nothing to do with math. Math, like any art form, is really about creating, imagining, finding new possibilities. But try telling that to nearly anyone who has learned about it through the American education system.

So when I start trying to describe how I made something, I often see looks of fear and panic. People back away slowly.

Which is really a shame, because how we create computer graphics is every bit as beautiful as the end result. And in many ways even more so.

Data-driven artificial intelligence is not going to replace human talent, but it is going to redirect it. Whether composed of pictures or words, the output of a General Purpose Transformer AI tends to be boring and generic, unless it is artfully steered by a skillful human.

Just as society gradually realized that photography can be great art — in the hands of the right person — society will eventually recognize the value of the work of talented people who understand how to direct AI to great effect.

But this means that future art will be different. A story will no longer need to be a linear sequence of events.

Instead, it can consist of characters interacting with each other in a semi-improvised way. A great novelist will shape the backgrounds and personalities of those characters, so that their interactions with one another over time are profound, or truthful, or just plain funny.

A story in the future may become something that you can revisit from time to time, like having lunch with an old and dear friend. You don’t expect to have the same conversation with your friend each time, but you do expect to encounter essentially the same person that you’ve always known.

AI will lead to something similar. Each time you read a future story, it might be a little different. Yet if well crafted, it will continue to thematically hold together, like an old and dear friend.

For the last few months I have been haunted by a math problem that I just could not solve. I needed to get it right for a computer graphics project that I’ve been working on.

I kept coming back to it, but to no avail. Eventually I put together a fake approximation which sort of works if you don’t look too closely.

Needless to say, that was not very satisfying.

Today, nestled at home on a Sunday afternoon, I decided to revisit the problem. I put in the laundry, had some homemade cold brew, ate a chocolate rugelach, and got to work.

And lo and behold, I actually solved it. Now I don’t need to use a cheap approximation, because I have the real thing!

I wonder why I was suddenly able to solve that problem today. Maybe it was the rugelach.

One day you will have a ChatGPMe (Chat Generative Pre-trained Me). It will be trained on all of the things you’ve ever said, written and done.

It won’t be intelligent, but it will serve as a faithful echo. When you are feeling lost, or experiencing bouts of self-doubt, you will be able to turn to ChatGPMe and ask “What would I do in this situation?”

And ChatGPMe will tell you, pretty much in your own voice. You will be able to access the best parts of you, those times in your life when you were feeling creative, confident, at one with the Universe.

Eventually, you will wonder how anybody ever managed to get through a day without ChatGPMe.

Do we have an organic mechanism in our brain that responds to beauty? Or is recognition of beauty something that we need to learn?

In other words, is recognition of beauty biological or cultural? It’s a difficult thing to test, because you can’t remove a biological human from his or her culture. Anything along these lines that you try to test will always be influenced by cultural influences upon the person you are testing.

One question we might ask is to what extent perception of beauty differs between cultures. If a person is considered beautiful in one culture, will they necessarily be considered beautiful in a different culture?

Something to think about.

It would not be an exaggeration to say that Jerome Bixby’s work traced the contours of my childhood. I remember vividly the day I first read It’s a Good Life, a story which can give me shivers to this day.

I also vividly remember watching the Star Trek episode Mirror Mirror, and wondering why it was so much better than the other episodes. And then, years later, stumbling upon The Man from Earth, so deceptively simple, yet profound.

Eventually I realized that these were all written by the same person.

Jerome Bixby would have turned 100 today. Let’s hear it for him, wherever he may be in the multiverse.

I first saw Jurassic Park when it came out in theaters in 1992. Like everyone else, I was in awe.

Every moment of it was masterful. Spielberg had somehow redefined the art of filmmaking, and we all knew it.

One moment in particular jumped out at me, in its sheer ornery cleverness. Seventeen minutes into the film, a tiny “blink and you’ve missed it” event reiterates the entire central conflict of the film.

It’s during the scene in the helicopter. Technophobic Dr. Grant, played by Sam Neill, can’t manage to fasten his seatbelt, and everyone is trying to help him to strap in.

At first he sits there helplessly, holding two female seatbelt ends. But then, in a moment of inspiration, he ties the two ends together, with a look of triumph on his face. So what was this really about?

You’re probably way ahead of me here. The central twist of the film is that the creators of the park had guarded again the possibility of dinosaurs breeding by raising only females. No males, no propagation of the species.

But because they had introduced frog DNA to fix the gaps in the recovered dinosaur genetic code, the dinosaurs gained the power to reproduce by parthenogenesis. No males required.

I other words, a seemingly tiny throwaway moment early In the movie foreshadows the giant plot twist that comes to dominate not just that film, but all the sequels that followed.

Perfect, simply perfect. This is one reason I go to the movies.

In a letter written in 1657, Blaise Pascal famously said “I have only made this letter longer because I have not had the time to make it shorter.” There is a deep truth in this seemingly ironic statement.

When you are trying to convey your ideas to people, it can be fun to go on at great length. But throwing the kitchen sink at the problem can easily become a form of mental laziness.

To condense your message down to a few well chosen sentences — that is the greater challenge. If you find that you cannot describe what you want to say in a short and clear abstract, maybe you still have more work to do.

Joseph Weisenbaum, who was born on January 8, 1923, would have been 100 years old today. If you’ve heard of him, it’s probably because of his ELIZA program — the very first chatbot.

But I think his greatest contribution was his well reasoned critique of artificial intelligence. For him, it wasn’t a question of whether we can implement ever more powerful AI capability, but whether we should.

Weisenbaum made a fundamental distinction between “deciding” and “choosing”. An AI can decide many things, only a human has the moral imperative to choose between what is right and wrong.

If we create an AI and have it decide things for us, the moral underpinnings of those decisions are still based on the human choices — and therefore the moral values — that led to those decisions.

Given today’s ever more powerful achievements in artificial intelligence, these questions are more important than ever. In the end, as Weisenbaum observed more than half a century ago, moral responsibility lies not with our AIs, but with ourselves.

My colleagues in Korea tell me that things there are now pretty much post-money. If you want to buy anything, or get swiped into your office building, you need your phone.

I asked one colleague what happens if you lose your phone. He got a panicked look on his face, and told me that it can take weeks of paperwork to re-establish that you are really you and get another authorized phone.

We are starting to get a bit of this in NYC, in our primitive way. I was one of the last holdouts still using a MetroCard. Now I just hold my credit card over the turnstiles like everyone else I know.

I wonder how all this will play out when those future glasses arrive, and replace our smartphones. Will the entire concept of money as we know it begin to fade away, when the process of checking out becomes a fully automated part of your shopping experience?

Wherever you go, and whatever you do, you will be continually charged as a sort of background tax on your lifestyle. Children will grow up having the instinctive understanding that certain activities drain your account more quickly than others.

Money, as we now know it, will be one of those fondly remembered things of old, like ice delivery or like having a telephone in your kitchen.