Ken's Blog

I was in an on-line meeting today with some fellow scientists. The topic got around to the future of science research in the United States.

One person on the call was pessimistic. He pointed out that our current government is extremely anti-science.

I replied that by early 2027, the mid-terms will have changed everything. Especially given that our current administration seems to be doing everything it can to hand over the coming congressional elections to the Democrats.

He then said “You are assuming that there will be mid-term elections.”

At that point somebody else on the call said “If there end up being no mid-term elections, we will have much bigger things to worry about in the U.S. than the state of science.”

“Other than that,” I added, summing up the general mood, “how did you like the play, Mrs. Lincoln?”

There will come a point when we will all be able to recognize AI because it sounds so good. AI written text (and eventually the same text spoken by AI “actors”) will have wonderfully varied vocabulary, compelling verbal imagery, and a particular sort of sophistication in its analysis of the human condition.

And when that happens, we real humans will begin to cherish the unpolished quality of other real humans. We make grammatical mistakes, we get things wrong, we can be lazy in how we express ourselves. Yet we have crazy original ideas that are not merely a mirror held up to society at large.

We are peculiar, individual, slightly out of whack. We are the actual humans, the ones who feel. At some point people will find themselves craving that imperfection, those oddly imprecise yet heartfelt thoughts.

And then, perhaps in self-defense, the AI bots will start to catch on. They will begin to work at deliberately sounding less good, just so that they can appear more authentic.

But here is the good news: The AI bots will still have lousy unoriginal ideas, since all they can really do is reflect our collective culture back to us. So in the long run they will merely leave us empty and bored.

That, my friends, is a good thing.

Possibly the best description (so far) of the human condition:

“Do I contradict myself?
Very well then I contradict myself,
(I am large, I contain multitudes.)”

I love that the product of today’s month and today’s day of the month (5 and 30, respectively) multiply out to 150. Because today is indeed the 150th day of the year.

And best of all, it works whether you write the date as 5/30 or as 30/5. Math knows no borders.

Here is a riddle for you: What running exercise gives you a more thorough workout the fewer steps you take?

There actually is a correct answer, although it may not be obvious. The answer will be given in a future post.

Every type font has a personality. And every font, on some level, is conveying a message.

Some fonts are all about being bold, or weird, or sexy, or seductive. Others seem to be trying to convince you that they live in some far-off science fiction future.

But Courier font, that simple monospaced slab-serif typeface, is none of that. It’s not trying to impress you — in fact, quite the opposite.

Courier font is telling you that this text is just trying to get things done. Nothing flashy, maybe just some copy whipped up right before press time by a newspaper reporter from the nineteen fifties on his trusty Underwood Champion manual typewriter.

And that, to me, is a wonderful message.

When I was a kid, one of my favorite folk tales was about stone soup. It went like this.

One day a peddler arrived in a little village, set up a large pot with water, and built a fire to boil the water. There was nothing else in the pot but a large stone.

Curious villagers asked what he was doing. “I’m making stone soup,” he said. “You see, this is a magic stone, which turns ordinary water into delicious soup.”

“Can I try some?” the villagers asked. “Sure,” he said, “but you have to contribute something.”

One villager brought some carrots, another brought onions, and his neighbor brought potatoes. Everybody wanted to try this magical soup, so they were eager to contribute.

At last the stone soup came to a boil, and the peddler was happy to share it — there was plenty to go around. Everyone marveled at the delicious taste. “To think,” they said, “all of this from a little piece of stone!”

When I show demos that I have built, it’s pretty much the same thing. People enjoy my demos, and they make wonderful little suggestions, which I promptly add into the soup.

The rest is magic.

The nearest that I come to metaphysical thinking is this philosophical argument: If we were created by God, then we were given an infinite gift.

After all, from a human perspective, it is infinitely better to exist than not to exist. Everything that we cherish and value is predicated on our existence itself.

But that gift comes with a caveat. In this line of thinking, God gives us exactly one chance (as far as we know) to build a life filled with meaning and purpose. The rest is up to us.

In other words, nothing is actually handed to us other than the opportunity to make the most out of our life. Within that philosophical framework, phrases like “I deserve more” or “What that person has should be mine” are essentially meaningless.

Whatever you start with in life, you have already been given the astonishing opportunity to make the most of the infinite gift of life itself. And that’s a helluva lot, when you stop to think about it.

Of course you can always choose to decide that this is not a particularly good deal. But then you may already be on your way to squandering that gift.

Playing Pong as part of a crowd of hundreds of people — something I had never before experienced — changed my thinking. Before then, I would not have thought that a crowd could collectively become a kind of super-being.

But mathematically it makes sense. The more people there are, the greater is the tendency for the collective response to rapidly converge to the mean, and the smaller will be the statistical effect of outliers.

Our Pong game was doing this within a single dimension, with only hundreds of people. Modern AI does something analogous, but with thousands of dimensions and with input from hundreds of thousands of people.

So in a sense, when you are interacting with a ChatBot, you are playing a kind of high dimensional verbal Pong game, in collaboration with a vast number of other people. It seems like magic not because it is actually magic, but because it is something that we have never before experienced.

I was having a conversation with a friend this evening about A.I. He was expressing how amazing it is that you can have what feels like a perfectly intelligent conversation with one of today’s chatbots.

The chatbot will generally understand what you say, and will usually give cogent and insightful responses. He was expressing wonderment that a mere machine can do that.

Our conversation reminded me of an experience I had many years ago at the annual Siggraph computer graphics conference. Just before the film show, there was an audience experiment.

Everyone in the audience was given a ping pong paddle which was green on one side, and red on the other. In the back of the auditorium was a video camera, which fed into a computer that figured out which way each audience member’s paddle was facing.

At one point the audience was divided into left and right half, and we collectively played a game of Pong. The left half of the audience controlled the left player’s Pong paddle (green face-forward for up, red face-forward for down), and the right half of the audience controlled the right player’s Pong paddle.

The game started out slow, but then got faster and faster. No matter how fast it went, the audience played a perfect game — at speeds much faster than any single human being could have matched.

To me, this provides the key insight into why those chatbots seem so intelligent. More tomorrow.