The Intelligence is in Application

These aren’t the AIs we’re looking for

AIs Remain Imaginary

As much as we talk about AI, it’s an open secret that it doesn’t exist. There is no single machine that the has the cognition and growth to meet the original ambitions of early futurists.

Here’s the bias: we humans like to think of intelligence in human terms. So it doesn’t matter if computers are somewhere between cool and awesome. We want a brain intelligence — literally a machine that mimics brain function and perhaps even development.

AI How to AI Why

But what if we black box the brain part? What if at a system level the input of information to the output of a decision or action was the determination. Isn’t the promise of exceptional results the reason why we are chasing the white AI rabbit?

How much does the ‘intellingence’ of the blackbox matter?

This removes the trap of thinking about the AI engine and more about the (often elusive) impact of AI application. If we look at what computers actually accomplish, the intelligence moniker fits much more easily.

Any sufficiently advanced technology is indistinguishable from magic. — Arthur C. Clarke

Tale of Two AIs

Even without the strict definition of AI, computers are performing tasks that were unthinkable by our meat brains even a generation ago.

Such as IBM’s Watson.

Who to avoid when playing Jeopardy.

Watson still needs supervision (translation: requires human chambermaid) as it learns. It still has a static source code. It is not strict definition AI.

Yet nobody can argue with the results. Incredibly diverse data is filtered, sorted and mined for trends at a scale no brain can compete with. Bunch of articles in, insight out. Sounds pretty intelligent.

DoNotPay only calls itself AI.

Go git ‘em

It’s a rudimentary chatbot that fights parking tickets. The actual App is incredibly simple — far from the lofty heights of a Watson and certainly not intelligent, right? Yet the ticket-slaying results are what I would consider borderline magical technology. Maybe even intelligent if you squint your eyes enough.

I would argue more people may be indebted to the application of DoNotPay than the engine of Watson. Which AI deserves to be called AI? Is application a more ‘intelligent’ implementation, like some sort of computer emotional intelligence? There is no right answer, but it helps adjust the bias that we evaluate the quality of AIs.

Intelligence, like many qualified features, is generally judged from outside observers looking in. Maybe even from a place where hatred of parking tickets needs — no, deserves — more champions.

Now I know AI is horrifically overused. As a self-confessed data nerd I roll my eyes as AI is used as a buzzword to make something mundane sound way better than reality. But maybe we should be little more gentle when someone gets a piece of technology working that makes a difference in society.

They Can’t All be Winners

Successful Application Requires Intelligence

The whole thing is useless without the pipes coming off the bottom.

Today, advanced computing is more available than ever (side note: LexPredict open sources itself- fellow data nerds rejoice!). Unfortunately, the current generation of AI is too often a rocket in a wind tunnel all revved up with no place to go. The missing secret ingredient to perceived intelligence is meaningful application of that raw power.

Application is problematic. Ask any tech entrepreneur. But there are big wins for anyone who can make it work. And bigger wins for larger society receiving the promised benefits of AI.