What Comes After Digital

Before we had digital electronic devices to play music, record our faces and voices, and show pictures and video, we had various media: photographs, phonographs, cassette tape recorders, film projectors, motion picture cameras, etc.

The digital transformation of the world has been remarkably swift and pervasive, for good and ill. We've never had so many movies and songs at our disposal, thousands of the latter easily contained in a single smartphone's memory banks along with thousands of pics, selfies, and videos. And nefarious governments can hoover them all up, stream them from surveillance sensors masquerading as consumer security devices, and the like. Yay.

I was recently pondering A World Remade by AI and realized that there's no reason anyone should stop at writing software.

As a brief aside, we still don't know whether intelligence can be implemented with digital computers. By that, I don't mean that intelligence is somehow only a biological thing.

To illustrate, consider the wind. An instrument that measures the speed and direction of wind (or any current of gas) is called an anemometer. So imagine that you had a 1-by-1 meter grid of these in a field, hundreds and hundreds of them (what we might call "internet of things" today) transmitting their instantaneous measurement of wind velocity every half second.

You could plot that grid on a computer screen to create one of those arrow drawings overlaying an aerial image of the field and watch the "wind" as represented by those arrows.

But that would not be the wind. It would be a visualization of the wind.

You could also take all those readings and create a "model" that, given some initial conditions, then computed changes to all those values at regular intervals, and then check how closely they matched the actual values. Someone might use something called "machine learning" or "artificial intelligence" to derive this model from the previous data.

When the model runs, it would be a "simulation" of the wind. But it would not be wind.

Tying this back to the question of intelligence, what if we can only create a model or simulation of intelligence, but not an "implementation" of intelligence, with digital computers?

With that in mind, consider the original question: what comes after digital? We had analog, then we had digital. It's much more compact and extremely flexible.

I suspect, the next thing is analog again. Imagine that instead of playing an MP3 (i.e. digital recording) of your favorite artist, an actual "artificial intelligence" of a sort could drive the speakers with an analog signal that was much like an actual artist using their vocal cords to vibrate air.

Or if you have driven a car with an autonomous driving capability and noticed the uncanny valley of inputs, what if the actual autonomous car was generating those truly intelligently as a human would, with all the subtlety of the terrain, the car, the road, the occupants, etc.

I'm sure there are other examples that might come to mind.

If we are able to create artificial intelligence, I have a hunch (that could be completely wrong) that it won't be digital and it won't be primarily over there writing software or summarizing our emails or any of the things people are attempting to do with LLMs today.

Just a hunch.