Month: January 2025

I just watched Yeon Sang-ho’s masterful film Train to Busan. I don’t usually like zombie flicks, but everything about this movie was brilliant.

What most surprised me was how much the character of Yon-suk, played by Kim Eui-sung, reminded me of a certain recently elected president.

I wonder whether that was intentional.

If robots are ever going to be a daily part of our lives, they will need to react quickly and accurately to what is going on around them. That includes responding to what people do.

There is no mathematical algorithm that will tell a robot what to do in a social situation. The only “correct” answer is whatever is the accepted social norm.

Since robots are not actually intelligent in the human sense, the only way this can work is via data driven algorithms. In other words, we need to use the same mechanisms of Convolutional Neural Nets and Transformers that have led to the rise of ChatGPT, MidJourney and their various algorithmic cousins.

And like humans, those robots will need to respond immediately, without needing to stop and think. This will be particularly true in situations that involve physical contact with humans, such as accepting or handing someone a cup filled with hot coffee.

Which means that we are going to need to develop robots that have machine learning (ML) embedded in those parts of the robot which involve interaction with the physical world — such as robot hands and fingers.

Eventually we will figure out ways to gather human data to train such robotic ML algorithms. But that won’t happen in isolation. We will also need more sophiticated touch sensors, proximity sensors, and other equivalents of the fabulous sensing machinery built into every human body.

So this won’t happen right away. But after it does, it will all seem perfectly natural.

Here is something that people do every day: They take a set of keys ouf of their pocket, juggle the keys so that they are holding just the right one, and then let themselves into their house or office.

The more you think about all that is involved in that process, the more impressive it is. And yet we do it without conscious thought.

It’s hard for robots to do something like that. To make it easier for them, I think we would need to focus more on the part where the entire action is done “without conscious thought.”

After all, we can only manipulate our keys this way because our brain is delegating most of the work to lower level thought processes, well below the level of consciousness. If we needed to actually think through every step of this action, we would probably never succeed in getting into our house or office.

How can we rethink robot intelligence to make them better at such “unconscious” tasks? More tomorrow.

When I was a little kid I loved to go through the dictionary. I generally skipped those little dictionaries they make just for school children, and went straight for the big grown-up one.

It was fun just to turn the pages, go through the alphabet, and see what I might find. Usually I was on the hunt for new words.

I was particularly drawn to words that seemed exotic, or odd sounding, or just plain surprising. Some words didn’t turn out to mean what I thought they would, whereas other words seemed to fit their meanings perfectly.

Sometimes I would spend hours just looking for really great words, learning what they meant, and how they might be used in a sentence. The dictionary turned out to be a vast and endlessly fascinating world.

We now live in a digital age where dictionaries are on-line. Kids don’t need to deal with old fashioned physical books, since meanings and usages can now just pop right up on their screens.

But that means that they aren’t exposed to the kind of random exploration which I enjoyed as a kid. They may never get a chance to experience the delightfully arbitrary juxtaposition of words that just happen to be alphabetically near each other.

Our current digital age is wonderful and empowering — we are traveling on an exciting road to the future. But sometimes I feel sad about the things that might have gotten lost along the way.

Here’s a fun game: On any given day, look at all the personages in history who were born on that day of the year. You can go as far back in time as you want, and you can consider candidates from all cultures and backgrounds.

Then you get to choose one that you would like to have dinner with.

Now here’s another game. Suppose that, for every person you know, you were shown a list of who that person would choose each day as their birthday dinner companion. But all of the lists are anonymous — you are not told which list belongs to which person.

The goal of the game is to see whether you can figure out, just from looking at those lists, which list goes with which person.

By the way, of all the notable personages born on this day in history — January 17 — the one that I would most like to have dinner with is Benjamin Franklin. Who would you choose?

I know that the world-building of Minecraft is based on my noise function. All of those cool mountains and hills and landscapes are created out of procedural noise.

To create all of those topographies you just need to use noise over a 2D domain. But the noise function is actually defined over a 3D domain.

I wonder whether it might be interesting to create a variant of Minecraft that exists not over the 2D domain of a landscape, but rather the 3D domain of space. Of course the gameplay would need to be different, and working out that gameplay would be an interesting exercise.

It seems to me that such a game would be particularly well suited to virtual or blended reality. The players could then more easily look around and change their viewpoint.

I wonder whether anyone has built such a game. If not, I might just have to.

Like millions of others, I have been watching the confirmation hearings with open-mouthed astonishment. It looks as though we will get an election denier as our next attorney general. And the pick for secretary of defense is a bizarrely unqualified boob (who admittedly has nice hair).

Since I have also been bingeing The Office, it is hard not to make comparisons. But here’s the thing. When you watch the insane clown show, and see all of those senators trying hard to maintain a straight face, you begin to see the characters in The Office differently.

At the end of the day, Michael Scott and his co-workers have a certain measure of dignity. Underneath the absurdist comedy, you end up seeing real people, flawed yet striving to be better.

That particular quality — the possibility of redemption — is one of the defining traits of absurdist comedy. But there is nothing at all redemptive about the current round of confirmation hearings. They are simply sad and tragic and stupid, although most definitely absurd.

Maybe in these hearings we are witnessing the birth of a new kind of reality television — absurdist tragedy.

As generative AI continues to improve, more and more of our on-line communication might be given over to the virtual version of ourselves. At first it will mainly be business correspondence, but then we might begin using our Chat Avatar more and more often for personal communication.

Eventually, the version of us that people know on-line may not be the live version of us, but rather a sophisticated proxy. This proxy will have been trained on the history of our communication style, our sense of humor, our likes and dislikes, and our various quirks of personality, endearing or otherwise.

But what happens when we die? Perhaps death will start to become a private matter, something not spoken of in polite company, or even acknowledged. After all, we will still be able to communicate with the world perfectly well, even after we are no longer here.

Maybe, in an extreme scenario, some devastating future event will wipe out all of humanity, but nothing will really change. We will continue to communicate and chat away with each other on-line, just as we had been doing for years and years.

It won’t actually be us. But you wouldn’t be able to tell the difference.

I’ve recently been bingeing The Office. Specifically, I’m watching the Superfan version, which I find vastly more interesting and engaging than the original edit.

To me the most astonishing character is Dwight Schrute. There is something so pure about him, so absolutely unsullied by the constraints of what the rest of us call “the real world”. Dwight is essentially making up his own alternate reality as he goes along.

In the Superfan version of the series, you really get to see inside his unique mind. The man is clearly somewhere on the spectrum, with odd gaps in his perception of others, but that often ends up functioning as a kind of super power.

While watching the show this time around, something has been nagging at me. Dwight Schrute reminds me of somebody, but I’ve been having trouble figuring out who.

And then just today it hit me. You can see it in the way Dwight thinks, in the way he relates to others (or doesn’t), and in the way he sincerely believes that reality, and all the people in it, are merely constructs for him to bend to his will.

Basically, Dwight Schrute is Elon Musk.

I have a theory about the difference in attitude about household robots between East and West. To recap, people in the West seem worried, and their minds can wander toward thoughts of monsters, whereas people in the East generally seem delighted by the prospect.

My theory is that it’s a difference between Judeo-Christianity and Animism. In the West, life is God-given, and to pervert that natural order of things can seem monstrous.

That is the basic premise of the very first science fiction story — Mary Shelley’s Frankenstein. Man using technology to compete with God is an act of hubris, and that act must be met with divine punishment.

In contrast, Animism is premised on the fundamental belief that there is life in all things, whether a tree or a rock or a chair. So to bring an inanimate object to life is simply to manifest something that was already there.