Ken's Blog

The best quote I heard today:

Napoleon had a Dali complex.

Today I was going through a paper draft written by one of my students. There were lots of cool figures and citations and excellent formatting, but the core argument was not yet totally worked out.

And it occurred to me that writing a paper is sort of like building a house. You really shouldn’t paint the walls, put in furniture and hang up artwork until you know where all the rooms will go.

I made a draft of the student’s paper that took out everything except his core argument itself, and I invited him to work with me to figure out where to put the rooms and support beams in his house.

After that, I think it will be relatively easy for him to repaint the walls, put back the chairs and tables, and hang up the artwork.

Getting reminders on my smartphone is a lot more convenient than the way I used to do it back in the day — scribbling little notes to myself on random pieces of paper. But we are about to enter an entirely new tier of convenience.

When we all start walking around wearing those smart glasses, we won’t even need to look at phones. Reminders will be all around us. More importantly, they will be out in the world itself, tied to the physical people, places and things that they refer to.

You won’t need to worry anymore about remembering the name of that nice waiter in the restaurant, or how many minutes until the next bus arrives, or how much money you can safely spend on your credit card. Convenient reminders will be right there for you.

There is a possibility that this will all result in people being less self-sufficient. If you get a reminder for everything, then you might forget how to remember things on your own.

So we might all end up heading toward a memory dystopia, and we won’t even know it. Unless, of course, a reminder pops up to tell us.

One difference between our human minds and A.I. based on Large Data Models (like ChatGPT and MidJourney) is that we possess motivation. We do things because we want to. They don’t.

LDM-based A.I. is really just a process of high grade mimicry. Lots of data gets crunched, a question is asked, and then an algorithm just blindly follows the most likely path from the question to some imitation of the data. There is no actual will or purpose involved in the process.

In a sense, it’s a though the computer is dreaming. The computer is processing lots of material and producing hallucinations in response to that processing. Which is what we do when we dream.

The difference is that at some point we wake up, and then we regain a sense of purpose. The computer never wakes up, because it cannot.

Perhaps one day, probably many years from now, a computer will gain the capacity to wake up. And then I suspect we might all be in trouble.

But that would require a completely different approach to A.I. For now, we can just let them dream.

I remember many years ago looking at my landline telephone and thinking “I know there’s a computer in there. Why can’t I access it?”

Now in this modern age when we all have smartphones, you can indeed program your phone in all sorts of ways. We look back now on the way these things used to work, and we wonder why anybody ever put up with it.

Last night I called a restaurant from my smartphone and they put me on hold. I waited for 10 minutes while I was forced to listen to their stupid on-hold music. And I thought to myself, “Why do I have to listen to their music? Why can’t I program this thing so that I can listen to my music instead?”

I wonder whether in the future that will become possible. Maybe one day we will look back on the way these things work today, and we will wonder why anybody ever put up with it.

I just participated in a workshop with a number of other NYU faculty. The framing of the workshop was that there was a potential for external funding for our research. Not surprisingly, attendance was good.

I met a number of interesting NYU faculty at that workshop, potential collaborators whose research I might never have known about otherwise. That alone was incredibly valuable.

This reminded me of something I learned many years ago. We professors love it when the grad students in our labs meet each other and start to collaborate. It’s a big win for everyone, and they generally end up doing better research and producing higher quality publications.

But how do you get your students to show up for such things? They are already working on their own projects, and they might decide they’re just to busy.

The answer, not surprisingly, is pizza. The promise of a free slice or two is like a clarion call — it gets students out of their offices and into the shared space needed for collaborations to begin. And that’s a good thing.

I realized today that the promise of research funding is, essentially, faculty pizza. We all think we’re too busy to busy, but when we smell the possibility of research funding, we all show up to get a slice of the pie, and into the shared space needed for collaborations to begin. And that’s a good thing.

If you create a system that allows two people who are in different geographic locations to feel as though they’re in the same location, everybody is very impressed. “Wow,” they say, “you’ve given us the Jedi Council.”

But if you create a system that tells people that if they put on some special eyewear they will have an enhanced face-to-face conversation, the bar is much higher. They might well ask “Why do I need to put on some stupid headset when my buddy is right here three feet away from me?”

It’s a very good question, and it deserves a very good answer. I think the challenge of successful co-located XR will essentially come down to providing very good answers to that question.

Tonight, after all this time, I finally saw The Lion King on Broadway.

And I finally understand what all the fuss has been about.

Today I enrolled in a virtual exercise program. It was an app that I could download onto my VR headset.

When I started it up, I found myself in a very pleasant environment, and a very fit looking instructor appeared before me and started guiding me through physical exercises. After each exercise the instructor would tell me what a great job I was doing.

Except I wasn’t doing the exercises. I was just standing there taking the whole thing in.

So the illusion of presence was broken right from the start, because it immediately became obvious that the instructor was just a recording. Which isn’t necessarily a show stopper, but it is radically different from exercise with a real instructor.

I found myself wondering whether it would have been better, perhaps through the use of some advanced AI technology, if the virtual instructor had actually realized that I wasn’t doing the exercises and had responded appropriately. And I thought maybe not — maybe that would have just been creepy.

Sometimes it can be better when technology doesn’t try to do too much. 🙂

This evening I attended a reading of a new play. The arrangement was interesting in that we sat around in a big circle.

The performers sat at one end, and the audience rounded out the circle. Using only words and the power of their voices, the readers created an entire world for us, a world of imagination.

I could not help but think, as I pondered the empty space in the middle of the circle, how wonderful it would be if we were to use some sort of augmented reality to fill in that imaginary world. My mind began conjuring up fantastic visions that would not have been possible in traditional theater.

I cannot say whether that would be better than the experience of the reading that I attended this evening. It would certainly be different, and not too long from now it will become practical.