How Will Your Children Live With AI?

I've read numerous times that 40% of the jobs in existence today didn't exist as little as 10 years ago. I have no clue how correct the numbers are, but clearly the career landscape changes on a constant basis. Every year the changes are getting faster and more profound.

For as long as robots have existed, the concern about their taking the jobs of humans has been raised as a key issue with automation. And, to a large degree, robots have displaced human labour in roles that are repetitive in nature. Take a look at any automobile factory or Amazon fulfilment center and you will see evidence of this. And although human labour has been displaced, somehow the market seems to keep creating more new roles and opportunities.

While we haven't had the mass unemployment that was originally predicted as a result of automation, we clearly do have wage stagnation. For the first time in modern history, a new generation is looking at a life financially worse off than their parents. Is automation to blame? I have no clue, but I am sure that I wouldn't have any problem finding experts who agree with me no matter where I stood on the topic.

As of late, the boogeyman has changed from Robotics to Artificial Intelligence. AI has lived in the shadows of computing for decades, always the domain of Computer Scientists and Hollywood script writers. In the last few years, however, AI has started to make itself felt, even if those who interact with AI are unaware of it.

When installing Windows 10 on a re-purposed computer only a few weeks ago, Cortana spoke to me. Having been tortured by occasional industry attempts at speech recognition over the last 20 years I was ready to start looking for a way to turn it off, but I gave it a shot first. Much to my amazement, Cortana easily understood me speaking German with an American accent. It did far better than I had thought possible and forced me to realise that the age of speech recognition, which is enabled by AI, has finally arrived.

A colleague recently shared her unexpected experience with AI and Natural Language Processing, when one of her clients said that his "assistant" would be in touch to arrange for their next meeting. The "assistant" did email my colleague and they quickly arranged for the desired follow up meeting. Only afterward did my colleague learn that the "assistant" was a bot. It did such a convincing job of mimicking a human, that it was indiscernible from the real thing. Wasn't that supposed to be the Turing Test?

Its clear that people already prefer machines to humans. Just walk into a supermarket and you'll see people line up for automated checkout stands while staffed registers are completely open. This is because bots:

• Are always polite
• Respond quickly
• Never forget their commitments
• Always read your *full* email before responding
• Have infinite patience (when was the last time your Navi complained about you missing a turn for the 4th time?)
• Will always try to understand you, rather than force their will upon you
• Will communicate with you at your own level and never look down on you
• and much more...

Unlike with robots, which require that a relatively expensive and complex device is manufactured, trained and maintained, AI implementations, to a large degree, deploy on existing infrastructure & via the Cloud. That is, deployment is very rapid, maybe even viral. Every new user is also a new teacher. Every interaction is an opportunity to learn and improve. When you have users numbering in the 10s of Millions or even Billions, the speed and breadth of this feedback loop becomes simply breathtaking.

AI is a train that is approaching at a speed faster then anything else we have ever experienced before. While robots took on repetitive mechanical roles, AI is taking on thinking roles. It has already started in legal, law enforcement, accounting, engineering, psychology, sales, medicine and customer service. With the speed at which AI is advancing, I think it is fair to say that precious little of what my kids are learning in school has much of a concrete application to their future. Considering that we are fundamentally visual creatures and learn best by doing, I wonder if we will even need to be able to read in 20 years.

To the Engineer who says that there is no way a machine can do his / her job, I suggest you extrapolate the rate of change over the last five years into the next 10. Take a close look at TensorFlow, CNTK, and AWS Deep Learning AMI. Change is getting faster, not slower. The changes are becoming more profound, not less. We either learn how to ride the train or we get run over by it.

And that's the problem. Socially speaking, we humans usually have to get hit on the head before we learn the value of wearing a helmet. The risks in our environments are so numerous that if we were to take precautions against each and every one, we'd quickly be paralysed by an endless stream of defensive actions. As a result, we learn to ignore hazards which we deem unlikely. Because only a tiny fraction of society really understands AI, it is largely ignored or taken for granted. Ignorance is bliss, in the truest sense.

Unfortunately AI has as much potential for evil as it does for good. You can be certain that the hackers who stole up to $1B between 2014 and 2016 are working hard on doing it again with AI, but this time adding a few zeros to the take. Rest assured that every half-way intelligent thief, con artist and fraudster is asking themselves how to add AI to their repertoire. Think dating scams and phishing via AI enabled bots. Think WannaCry, but only a lot more powerful.

And because we typically learn "after the fact", you can be assured that it will take a significant catastrophe to wake us up to the need for corrective action. The problem is that the potential power of AI is so great and increasing so rapidly, that we may not get a second chance. Even if we do get a second chance, we won't be able to predict with any certainty where the next shot is coming from or how to protect ourselves from it. We don't know where this train is going and it doesn't have a braking mechanism.

I mean, hell, we can't even agree on how to handle a global challenge that 99% of scientific community says is on our doorstep, much less address something as vague as an imminent disaster caused by an as of yet undefined and unknowable source.

The impact of AI on society is simply unpredictable. The potential scenarios run from dystopian to utopian, with human greed and short sighted self-interest tending to promote a Mad Max like outcome. One thing is certain - your kids and mine are going to have to navigate this and there is precious little meaningful advice that we can give them. Nothing that we have lived through can compare to what they will experience.

How are you preparing your kids for their future with AI?