The Artificial Promise of AI

AI Is Nice, But Is It Consequential?

I love intelligent assistance. I wrote earlier about my favorite features in Apple’s new iOS 10 which both involve learning how I operate. Having my iPhone automatically categorize my e-mail and suggest playlists is pretty cool. But lets get real about how earth shattering important this is. There’s a lot of hyperbole floating around right now about how AI is the next frontier in device evolution. Is it? If automating menial tasks is changing the world then I suppose the Roomba is leading the way.

I see intelligent assistance similar to home automation. It’s great automating your door locks and light switches because it’s a convenience and one less thing you have to remember to do. However, it’s not changing my world. Likewise, I fail to grasp the real importance behind AI on my phone in the future.  It seems like it would be nice, just not very consequential. Apple and Google are on the right track with the development of intelligent assistance. It’s a necessary next step. However, where I’m skeptical is just how important it is.
 
And further confounding the problem, Google is setting up AI to take a big fall. Google is about to do to artificial intelligence what Tesla has done for autonomous driving. Oversell something so egregiously that the public starts to see it as a negative. And why? Unlike Tesla, Google isn’t based on a pyramid scheme dependent on new money flowing in.
 
I work with data on the behalf of Fortune 500 level companies and I have some familiarity with what you can do with it and where it’s limitations are. Statistical analysis can help a great deal when trying to budget your resources based on past trends. But it can never, ever, predict the future. It just can’t and it never will. And the size of the data set doesn’t make a difference. It only crystallizes those trends. If large data sets could predict the future professional baseball players wouldn’t have to play the game and weather forecasts would be accurate.
 
So why is artificial intelligence fools gold? Let’s think about this logically shall we?
 

1. If it's repetitive, you’re less likely to need help with it – TechCrunch reported on a story of one of their employees who shut down Google Now because it started giving him directions to a bar that he would visit on some Sundays. He didn’t want any one to know about that. But here’s the thing, if you’ve already been there a few times, you don’t need help from AI. What would be helpful would be directions to some place that you’ve never been. But AI will never be able to  read your mind. It won’t know that you are in the mood to try something new on Sunday night. Where Google’s AI works best is in when you have established patterns. Sorry, but that’s useless for most people. The least likely thing you’re likely to need help with is something you already do.
2. It’s a glorified list maker – So let’s say that Google’s Assistant or Apple’s Siri reads all your e-mail, listens to all your conversations, and sifts through your web browsing history. In return, what would you get? Once in a while it will remind you to check in for a flight or get restaurant reservations. Boring routine type of stuff. How is this different from writing a reminder on a post it note? Is this not a glorified list maker? This is not world changing stuff. These are menial tasks that don’t mean much.
3. Even people misunderstand each other all the time – I’ve used this example before but it describes what I mean perfectly. A little while ago I told my wife that we needed to get coffee. She wasn’t sure if I meant go to Starbucks and relax outside or if we needed to run to the grocery store to stock up on supplies. These scenarios are the opposite of each other. If REAL human intelligence can’t get it right how can artificial intelligence ever hope to come close? AI will never be as good as your significant other at understanding what you really mean when you say something. And your significant other gets it wrong all the time. If people go berserk when their software applications have occasional bugs imagine how sick of AI they will be when it fails to understand what they said on a regular basis. And it won’t ever get better until machines can read our minds

 
I do believe that there is a place for intelligent assistance. However, I feel that Google is overplaying their hand and damaging the whole concept of artificial intelligence. They shouldn’t call their version of intelligent assistance “Artificial Intelligence” any more than Tesla should call their glorified cruise control “Auto Pilot”. They are setting sky high expectations for a very young technology.
 
They need to do the opposite and lower the bar to help manage expectations. Sooner or later people will realize it’s never going to live up to the hype and become disillusioned. Or worse people will try to depend on it too much and be let down when they really need it.
 
It’s funny to me that the word “artificial” is even part of the name. That’s typically a derogatory term to denote something which isn’t as good as the real thing. Who wants an artificial leather coat or diamond ring? In the case of artificial intelligence, real intelligence would be you doing your own thinking and planning or hiring a person who makes it their job.
 
So why is Google making such optimistic promises regarding AI’s future? It’s because they are asking for huge amounts of data and something needs to distract the consumer. But by the time the public figures out that AI is destined to always be a poor substitute for real intelligence it’ll be too late.