I don’t try to predict the future, but I do love how science fiction provides rich and detailed storylines for exploring “what if” scenarios. Eons from now, we could be battling an underground rival that’s hacking Earth’s central intelligence, or racing against time to defeat the reappearance of the great evil. In any case, I think the most fascinating part about futurism is how we get there: our relationship with technology.
Machines seem to have two major archetypes in sci-fi. The first is the “Trusty R2 Unit”—the sidekick, the good buddy. Here, technology plays a continuous helping hand in keeping us humans from doing grunt work. The other side is The Terminator version, where we humans build a superior intelligence that wakes up to destroy or enslave us.
I’m unsatisfied with just having technology as a buddy to take the day-to-day labor off my hands—I honestly love getting my hands dirty with plumbing or construction. But, of course, I have no interest in being wiped off the planet or used as a power source à la The Matrix. In a universe of infinite possibilities, it’s an exciting challenge trying to figure out how to make my vision of the future a little closer to reality.
So, what would my intelligent system look like?
- Simple but powerful command interface: These systems need to have more expressive interfaces and commands—by an order of magnitude. They need a compact and expressive set of commands that work from the top down, helping the human refine and explore the solution set. Imagine we had an intelligent system that provided a home-building architecture service for us. I would come in and say, “I want a modern house for a family of five,” and perhaps share a few online pictures with the system. I’d have clearly specified a solution space with no details and the machine could start there with that compact instruction. We’ve done a huge amount of work with 10 words.
- Immersive data navigation: Next, the system needs to provide an interactive interface where the human can visualize and explore the solution space. There are many options, many views of the data, and these things should be presented clearly and simply to the user. This must be an immersive experience offering the ability to pivot and change the perspective on the data, zoom into something interesting to get more data, and zoom out to see the big picture. It is effectively the reverse of the point above; now the intelligent system must be able to communicate in a clear and expressive way back to the user about all the terabytes of data it knows. We need to go beyond tables and two-dimensional charts. We need to find ways of using size, shape, color, opacity, and—maybe in the future—texture, weight, and so on, to communicate multiple dimensions of data perspectives, and allow the human to truly explore the data space. In building a house, you might want to see traffic patterns depending on where furniture is placed, or hot and cold spots, or energy consumption over a whole year. Think about how awesome it would be to make choices about your custom house and see the impact immediately on hundreds of different metrics that you prioritize in real time.
- Able to answer the question “why”: Running “what if” simulations is something computers excel at today, but truly intelligent systems will need to answer the “why” question for us. Knowing why things are the way they are makes us smarter and allows us to make better decisions (or at least provides that possibility). In this world, we would just say, “Build me a house,” and the intelligent system would fill in all the details to make it a reality. However, when you see the options, you’re going to end up asking, “Why did you recommend we paint that wall orange and that one wall red?” And you’re going to expect there was a set of logical reasons to the colors chosen, rather than the intelligent system just running an algorithm to choose three random RGB values for each of your rooms (I can’t even imagine how ugly that house would be).
I call this view of intelligent systems the “Data Commander” pattern (maybe I have watched too much Star Trek). It’s not one where we have a technology pet, and it’s not one where I am in the cage. It is a view where technology is more powerful, but where I can visualize and see further and deeper into the data of my world, asking “why” questions to guide things in better direction.
It is obvious that machines are tools, and sci-fi has scared us into thinking of the rise of the machine as the rise of an enemy. But this line of thinking, while not impossible, is lacking imagination. The real challenge is to think in the in-between space—building intelligent systems that are our true counterparts.