Two Predictions on AI

There are two things that I'm nearly 100% positive that we'll see in the years ahead when it comes to AI. It doesn't take a prophet to predict either of these because I am simply looking at current technological trends and extrapolating them forward.

The first is AI-based cyberattacks and hacking. Right now, nearly every hack or breach is initiated by a human being, though this will not be the case in the not-too-distant future. Eventually, we will become quite accustomed to the idea that "bots"--as they are currently referred--will act all on their own to steal identities, penetrate networks, and conduct all sorts of malicious activity. 

Keep in mind, this is already happening now since cleverly programmed software has the ability to propagate like a virus from host to host without the need for a human hacker to facilitate its movement. This is common knowledge to all of us, which is why we collectively speak of such malicious programs as viruses, which, like their biological cousins, inhabit that gray area between living and nonliving entities.

We now wrestle with this same ambiguity when it comes to nonbiological viruses. Due to the growing complexity and ability to act without human direction, i.e. autonomy, malicious software utilizing the latest advancements in machine learning will also begin to appear more and more like a living organism with intelligence similar to or greater than a human being.

I suspect that we are much further along in this than most of us realize and it is only a matter of time before we see headlines of an AI-based cyberattack originating in a government research laboratory.

The second prediction I am making--and this is not really a stretch--is that most of us will be having conversations with AI on a regular basis. I already do this on a limited level with Siri on my iPhone when I ask her questions about random facts that I need to lookup or when I need her to send a text message or set a reminder, etc etc, but as the programming gets better and those at Google, Apple, and Amazon apply the latest advances in NLP (natural language processing) and machine learning, the conversational quality of AI will eventually become so natural that it will seem like we are just talking to another human being. As many of you probably know, this is referred to as the Turing test.

The first scenario of an AI-based cyberattack could happen any day. The second scenario is probably within a matter of years. I think Ray Kurzweil has predicted the Turing test to be passed by 2029 but we will see and be quite comfortable with conversational AI long before that.

Then, when AI starts forming its own opinions and thoughts on matters we ask it, that's when things get interesting.