Ken's Blog

After the XR/AI singularity, when everyone is wearing those future glasses, your personal AI assistant will know a lot about all of the physical objects that you can see. I wonder how that knowledge will be used in practice.

For example, will you choose to add annotations to your morning coffee? When you look at your coffee cup, a neat little text might pop up just above it, giving you all sort of useful information.

How fresh were the beans? What is the actual blend, and how caffeinated is it? How hot is your coffee at the moment? And how much more is in the pot?

All of the objects in your lives could be similarly annotated. Where is that screwdriver, the one which fits the particular screw that you are looking at? How much would it cost to add another shelf to that wall hanging bookshelf, and how soon could it be installed?

Will your relationship to your physical objects change once you are able to query them? Or will you just get tired of the whole thing and take off those darned glasses?

Unless you are physically in the same room with someone, you can no longer know for certain that you are talking to a human being, rather than an AI proxy. An AI can do a pretty good job these days of imitating a person, at least for short conversations.

If an AI agent has been trained on years of someone’s emails and texts, it can create a plausible simulacrum of how that person speaks, and even what they might choose to say. This might eventually lead to some interesting situations.

For example, sometime in the future you might decide to take the day off, and let your trusty agent respond to annoying emails and texts in your stead. As AI improves, your little “vacations” might become progressively longer.

At some point, you might just decide to let your AI agent take over those annoying tasks entirely. Meanwhile, your colleagues may be having the same idea.

Eventually, all business correspondence might actually be conversations between AI agents.

I wonder whether anyone will notice.

It was 30 years ago today when IBM’s Deep Blue defeated Garry Kasparov in chess for the first time. That was the first shot across the bow.

Now it is 2026 and here we are. Who knew?

In yesterday’s post I talked about a potential downside to the integration of AI into the creative process. In short, relying on AI can lessen our understanding about how things really work, which can make it harder for us to create something truly new.

But there is another, more positive, side to the equation. Let’s put aside the quasi-religious nonsense about AI replacing human intelligence (or, in the case of OpenAI, the misleading marketing).

While AI doesn’t actually think — in the human sense of that word — it is a great translator. AI is good at absorbing and reproducing idioms and terminologies, which can make the expertise of a plumber or a doctor or a physicist more comprehensible to those outside of their respective fields.

The use of AI has the potential to lead to more fruitful and productive interdisciplinary conversations. And that can help humans become even better at one of the things we do best: Collaborating with our fellow humans.

When you build something from scratch, crafting every component yourself, it can take a long time to finish your product. It would be so much easier to start with ready-made components, from off the shelf, and just snap them together.

Also, given constraints on time and effort, you might get more polished results by taking the ready-made path. But if your goal is to understand how something works, to figure out new and generalizable principles, the “build from scratch” approach has definite advantages.

When you build something yourself, from the ground up, there are no black boxes. You understand how each component works, as well as how and why those components all fit together.

To me the great danger of the trend to use A.I. for everything is precisely this. A.I. will assemble things for you, often from high polished components.

But it won’t give you any understanding or insight. People will end up with shiny artifaces, and on the surface it will all seem great.

But the human advancement of knowledge and insight, the hard work of understanding, will continue to slow. And eventually we will have no idea how anything actually works, nor how to make the intuitive leap required to create something truly new.

The ugly “Lion King” social media attack on the Obamas by that fool in the White House was problematic for many reasons. It was puerile, it was racist, and it was yet one more sign of a limited, senescent and rapidly failing mind.

But there was another problem with it. It was based on the wrong Disney movie.

The real reason the fool has such animosity toward the Obamas is that they are everything he could never possibly be. They are thoughtful, intelligent, well read, sophisticated, and deeply humane.

His feeble and fading brain dimly understands that their superior state of being is something that he could never achieve, and this fills him with jealous rage. A more accurate expression of his emotion can be found in the lyrics of King Louie’s song from another Disney classic, “The Jungle Book”:

I wanna be just like you
I wanna walk like you
Talk like you too
You’ll see it’s true
An ape like me
Can learn to be human too

There are lots of things for which we used desktop or notebook computers. And there are lots of other things for which we use smartphones.

There is some overlap between the two, but there are also clear differences. Unlike that computer on our desk, the smaller portable device in our pocket is useful when we are out and about in the world.

We use it to navigate while driving, for looking things up during casual conversations, to send a quick text to our friends. And of course, to talk with our friends.

As Marshall McLuhan said “The medium is the message.” The smartphone is useful in so many ways precisely because of how it can blend into our physical reality.

I’ve been trying to think about the use cases that come next. Specifically, how will use cases change when that phone is largely replaced by those future glasses or contact lenses?

When that happens, the information that we need will no longer be on a little screen that we take out of our pocket — it will be transposed onto the physical world around us. For all practical purposes, it will be a part of that physical world around us.

I suspect that in terms of general utility out in the physical world, we can make an analogy: Your future glasses will be to your smartphone as your smartphone is to your desktop computer.

It’s amazing how much time most people I know (including myself) spend looking at screens. For much of work and play, these artificial digital rectangles have supplanted the physical world around us.

When we start wearing those future XR glasses, will that finally change? Will we collectively break free of our rectilinear chains, and go back to embracing the world around us?

Or will we all start staring at make-believe digital rectangles which just happening to be floating in the air? Gosh, I hope not.

I was meeting with colleagues today about a research project. We’ve just recently started getting our first experimental results.

The meeting was largely do decide what we should do next. We could go in many directions, but we had to focus and choose one.

There is always a temptation to immediately go large based on that first early data, to gather lots more data, to instrument up in grand style. But we realized that this would be a mistake.

If you try to figure out too many things at once, it not only takes a lot of time and effort, but you also end up creating more opportunities for confusion and error.

So we decided we would run some modest benchmark tests to figure out how much we could trust those first measurements. The results of those benchmarks will tell us how much of what we have so far is actual signal, and how much is just noise.

At the meeting, here is how I ended up describing the next thing we need to figure out: “What the hell are we looking at?”

Just from a practical perspective, I think those are seven very useful words in research.

Today I became curious about one of my favorite poems, Do not go gentle into that good night by Dylan Thomas. So I read up on it on-line.

I learned lots of things. For one thing, I learned that the poem is formally a “villanelle” – a 19-line rhyming poem with five ABA stanzas followed by one ABAA stanza, with the first and third lines of the first stanza repeated multiple times.

I also learned that Thomas wrote it to his dying father in 1947. Also that his father, who taught English literature in Wales, managed to hang on to life until 1951.

I also learned that the poem was not actually published until 1952. Now that I know all these things, I am not likely to forget them.

The reason I am mentioning this is that it is quite possible that some day that entire process of “reading up on it” will take place entirely in people’s heads, in a matter of moments, due to forthcoming advances in AI and other technologies. It will be the real-life version of Trinity in The Matrix learning to fly that helicopter.

I am not sure that this will be a good thing. After that transition, we may need to redefine what we mean by “learning”, and we may need to say goodbye to this slower paced reality, where you actually need to look something up to learn about it.

Am I ready for that brave new world of the future? To quote Trinity, “Not yet.”