The following is a reprint of an article one of the Polywise founders posted on LinkedIn in January, 2017, which begins to put the current moment in context.

For years, I’d been very proud to say that I spent much of my senior year in the Computer Science department at Princeton writing papers refuting artificial intelligence as a concept (in a class specifically ON artificial intelligence, no less). My central premise was that any software designed by humans would always be limited by the scope used to define it, and that the software could never make a creative leap, i.e. a connection between unrelated concepts, that wasn’t originally provided to it by the human programmers. In other words, the machines could never be smarter than us.

Of course, I was wrong.

In my defense, I wrote those papers over 20 years ago, and the scope of what was possible in terms of technology probably eluded even my vainglorious younger self. After all, I’d been told by a network engineer at Bell Atlantic two summers prior that no one would ever find a reason to need more than a gigabyte of memory (which, at the time in 1993, was roughly the size of a piece of living room furniture).

What I hadn’t considered was that we human beings ourselves don’t know the full scope of the parameters that we set. We don’t account for all of the variables and possibilities of our own problem sets, and we certainly don’t tabulate and measure them all according to our own set of metrics & KPIs to determine the best possible outcome. We don’t disregard our own pre-existing biases that flow in a contrary direction from the optimal solution.

Finally, I hadn’t imagined a future where memory, storage, and processing power was not only cheap, but is, for all intents and purposes, infinite, thanks to cloud computing and the disintermediation that’s happening in the hardware component market.

In short, even if you accept my premise that machines cannot exceed the parameters of their programmers, if you apply those parameters to the state of the art of artificial intelligence today, it still gives you the means to vastly improve upon the resolution of any complex problem set presented to humans in unprecedented timeframes for a fraction of the man hour costs that would otherwise be required. For instance:

• Cond Nast is now using IBM Watson to find the social media influencers who can be best paired with their brand sponsors for maximum effect.
• A team of research neurologists were able to use IBM Watson to find which among the 1500 genes in the human genome actually cause Lou Gehrig’s Disease.
• Facebook is actually using its own home-grown AI applications to create NEW AI applications to optimize its existing software stack.

This was all in my mind when I noticed that the business press had started to catch up with this movement, largely driven by the releases of some key AI platforms and APIs by some of the biggest name in tech.

But, to my mind, there are some major disconnects amidst all of this press. Developers may be excited by the technological advances heralded by these tools, but for business decision-makers, the implications for their bottom line may be lost because there’s often very little talk about the sort of problems these tools can solve and how they’re applicable to their day-to-day operations.

With that in mind, in addition to my usual missives on the state of media tech, I’m going to use this space to regularly write about the AI tools that are currently available, either as APIs, open source code, or standalone products, and how they’re either being applied or could be applied in an actual business context. It’s my hope that, by devoting some time to unpacking the business value of these tools, many corporate decision-makers can see both the opportunities as well as the paradigm shift required to harness their success to the wave that’s already coming.

As you can see from this graphic from the team at Bloomberg Beta, the space is exploding in terms of new vendors and innovation right now. To say that we’re only in the early stages implies that very little is happening right now. Of course, if you look at Google Translate or Facebook Messenger or any number of apps that live in your pocket right now, it should be clear that AI applications are becoming ubiquitous in ways that are almost invisible to the human eye.

But when I say that we’re in the early stages, I really mean that the scope of the software, and consequently, business transformation that we’re going to see over the next few years due to AI will be exponentially larger than what we’re seeing today. I would even say that we can’t even imagine yet just how different tomorrow will look.

At least, not yet.