Final projects

This evening I had the pleasure of seeing the final projects of the 42 students in my graduate computer graphics class. There was no set assignment for this final project — everybody was asked to build on what we learned during the semester to create something that was meaningful to them.

Of course some projects were better than others, as expected. But the very best projects were so awesome that they took my breath away.

At moments like this, I really enjoy being a teacher.

Prospectus

Our lab, in collaboration with another lab at NYU, is working on a project to be presented the ACM/Siggraph 2025 conference. It will be multi-participant blended reality, and it will be very cool.

One of the first things we need to do is create a presentation of the project. We need to do this before we actually implement the project itself.

Of course we need to do this as part of our planning and production. But we also need to do it to fund the project.

Essentially, this presentation is a prospectus that we shop around to various high tech companies. We invite them to give us money, and sometimes they say yes.

They don’t do this because they like to give money away, but rather because it is an accepted way to create a bridge to our academic research. And if all goes well, they will end up discovering which of our highly talented grad students they want to hire.

So everybody wins. Also, we all get to prototype the future. 🙂

TalkingGPT

After everyone is wearing those XR glasses, I wonder what effect it will have on everyday conversation. At some point a killer app will show up that gives you a real-time “cheat sheet” to become a more sparkling, witty and generally engaging conversant.

Will this new capability be openly embraced? Will people launch into conversations knowing that all parties are experiencing a sort of artificial enhancement?

And will that lead to a transformation of conversation as we know it? For example, given that everyone will have their own private prompt, will some people choose to speak entirely in rhyming iambic pentameter?

Will movies and theater change because the everyday speech they aim to mimic will itself have changed? Will the unenhanced speech in movies from an earlier era come to seem odd and quaint, the way we find the inter-titles in silent movies odd and quaint?

Alas, once we go there, I suspect that we will never be able to go back. By analogy — we live in a world where everyone wears shoes, and therefore we build sidewalks that are unkind to bare feet. Similarly, once real-time conversation becomes enhanced as a matter of course, the world will become unkind to anyone who tries to speak without it.

Automat, part 2

I found myself thinking about the Automat recently while I was using Midjourney. As you give it various prompts, Midjourney creates images for you suggested by those prompts.

However, I soon started to notice that I was seeing images which looked a lot like other images that had already come up. And eventually I realized that I was seeing the work of actual humans, refracted through the software.

These were not simply computer generated sketches, but a collage of images that would be easily recognizable if seen in their original human-created contexts. In a sense, any software system based on an LDM (Large Data Model), such as ChatGPT or Mid-journey, is a kind of Automat.

It feels like a set of shiny high-tech drawers rotating past you, and you get the magic feeling that you can simply put in your coin, reach in and pull out a fully formed sandwich, apparently made by sheer wizardry. But behind this bright shiny surface are all the little ladies that you never see on the other side of the machine, crafting those sandwiches.

The difference is that in this case the ladies don’t realize that they are crafting sandwiches for you. They never actually intended them to be for you in the first place.

Automat, part 1

My mom tells me that when she was young, the Automat was very popular. An Automat was a slowly revolving vending machine filled with glass covered drawers.

When you saw something you liked — a sandwich or soup or a nice dessert — you would put a coin into the machine and take out the item of your choice. The next time that particular revolving drawer came around, it would contain some other yummy treat for purchase.

Behind the scenes, ladies were working to make all the food and fill whatever drawers were empty. You never actually saw these ladies. To the consumer, it all just seemed like magic.

There is an interesting parallel here with modern A.I. More tomorrow.

Old friends

Yesterday I did an extra lecture for my computer graphics class. It consisted entirely of old favorites — various interactive things that I created using computer graphics stretching back literally decades.

The students seemed to really like it, and for me it was an interesting journey as well. I found myself remembering where I was when I created this or that, and what else was going on in my life at the time.

Revisiting my old computer graphics, and talking about them, was opening a portal into my long term memory. As the lecture progressed, people and places I had not thought about for years came to mind.

And perhaps more important, I realized that the computer graphics creations themselves were, in their own way, old friends. And some times it can be good to visit an old friend.

Glass houses

The more I think about the reaction of those students yesterday, the more I appreciate the way technology causes fundamental shifts in collective awareness. I grew up in a world where privacy was a paramount value. You had your private life and your public life, and you generally did not like them to mix.

But the ubiquity of phone-based apps and social media is gradually changing the equation. We may be approaching an age in which younger generations expect to live in metaphorical glass houses.

Where you go, what you purchase, who you are spending your time with — all of those once private aspects of life are becoming ever more available for public display.

The underlying reason for this, to put it bluntly, is that this setup is good for commerce. The more that consumers become influencers for each other, the easier it is to successfully advertise and sell things.

But that isn’t how the change is generally perceived. People don’t say “I’m doing this because it helps advertising.”

Rather, they accept it as a fundamental and pervasive shift in values. When convenience walks in the door, privacy goes out the window.

Pay by phone

Today in our lab at NYU I walked by as three grad students were talking. The topic was being able to pay for things with your phone.

One student was telling the other two “China is way ahead of us on this. For years now you can pay for everything with your phone.”

Another student replied “I’m sure we will get there.”

I couldn’t resist piping in. “The reason I like paper money is that nobody needs to know who I am.”

For a moment the three students just stared at me, trying to understand why that would be a thing.

And then I saw a dawn of realization in their eyes. It was because they were talking to an old guy!

Ad absurdum

In the future, after all professional writers have been replaced by chatbots, there may be a falling off in interest in reading on the part of humans, because everything you read will start to feel similar and predictable. Kind of like if every movie came from the Hallmark Channel.

The good news is that there will still an opportunity to sell advertising. After all, humans are not the only ones who read. Chatbots also need to read, because that is how they train.

We are approaching ever nearer to the negative loop that Jaron Lanier predicted 20 years ago — chatbots being trained largely on the output of other chatbots. Yet even in that dystopian future, ads will still be incorporated into the training of chatbots, to make sure that they promote commercial products.

This is good news for the advertising industry. Long after most people have given up on reading, ads will still be needed, which means you will still be able to make money in advertising.

Then again, it might be boring work. After all, you will mostly be spending your time training an AI to create those ads for you.