As we steadily emerge from what has been called the “AI winter” — a ‘no-man’s-land’ of stalled or delayed progress that the field of machine intelligence has been stuck in for the past 30 years — we are finally beginning to see the first sprouts of true artificial machine intelligence springing forth all around us. It’s all begun innocently enough in the form of phone apps and sweet-talking assistants — all helping us speedily navigate the dense thicket of knowledge that we are fully immersed in 24/7 nowadays, like it or not.
Impressive breakthroughs in what’s recently become known as “Narrow AI” (i.e., machine intelligence that is supremely focused on the performance of a narrow range of tasks, such as search engines, driver assisted braking, GPS navigation or auto-parking on some newer cars) has begun cropping up in all sorts of devices that we never knew needed to be so smart (or talkative!).
Such mid-level acts of computer brilliance first began appearing as “anticipation bots”, such as Google auto-complete… that annoying ‘hurry up’ drop menu you get while searching where Google kind of guesses what you’re looking for and begins to return results before your slow, clumsy organic carbon-unit fingers can finish typing. By the time you glance up, your likely result is already cued up for a click. The technical term for this educated guesswork is ‘affinity analysis’, and you can see it at work big time whenever you browse around sophisticated merchandising sites like Amazon. Watch how “page bots” operating invisibly in the background keep feeding you product ads based on the sorts of things that you personally have been looking at recently, or have actually purchased in the past.
They know what you have an AFFINITY for, you see. Whatever it is that gets your shopping instincts stimulated and your buying juices salivating!
This technology is going to be making stunning advances in the near future, as there is a tremendous economic incentive to have advertising so precisely targeted towards specific individuals. The AI managing your personal ad feed will be tasked with making a judgement about your emotional state to understand when the best moment to present you with some kind of offer is at hand.
It knows your weakest moment and is ready to exploit it!
This steady rise in computer aptitude continues to accelerate as we move forward into the 2020’s (futurist Ray Kurzweil has predicted the appearance of the ‘Singularity’ — that moment when artificial intelligence will begin to equal human intelligence in all areas of psychological measure — somewhere around 2029 to 2034. This is known as Artificial General Intelligence or AGI. Regardless of the exact date of arrival, this singularity moment is sneaking up on us right now. Soon AGI (human-like) machine intelligence will be able to understand the various subtle nuances of human communication — skills that currently lie somewhat beyond what an AI can easily do (though they are catching on fast!)
Things like facial recognition (a skill that human brains are extremely good at, almost from the moment of our birth). Or learning to interpret subtle distinctions in a person’s tone of voice, modulation that suggests whatever is being stated verbally is being said in jest, or being presented as sarcasm, etc. We humans pick up on these sorts of communication nuances instantly and frame them within the context of the overall discussion we are having with a particular person. These are high-level language skills that AI will steadily improve on as it moves towards AGI.
A crucial part of human communication is the ability to quickly develop a sense of the person that we are talking to: are they being trustworthy or lying to us? Are they on an intellectual peer level? Intuition bred of experience goes into such assessments. If it is ever to possess a human-level understanding of the world, an AI must be able to decode nuanced communication in a similar fashion. A lot of our own personal psychology is woven into the way in which we communicate — there’s usually more there than meets the ear!
An AGI must also be able to sort out all the potentially inaccurate and confusing pronouns that people use when they speak. People can easily follow the context of a discussion, even though it may be unclear in a technical sense what precisely we are talking about:
“Hey Joe, I was wondering if your wife and that other girl that she was with the other day over at that place we were at, remember… where that one guy there was acting like a jerk? I was wondering if she could give her a call and let her know that I was thinking about asking her out this Saturday? I would really appreciate it if you and the missus could set me up with her, I thought she was really cool…”
You and I can follow conversations like this with ease, but how does a machine separate the genuine communication “wheat” from the verbal “chafe” that humans typically bury every point they make in? Right now, with great difficulty, but they are getting better and better at it — driven by innovations in speech recognition such as Siri and Alexa.
A true Artificial General Intelligence will need to perform with the same mental dexterity that a young child possesses after only a few years of hanging out with his or her parents. Of course, that we are not there yet, but researchers are working on machine intelligence relentlessly. As AI steadily approaches full sentience, it will be making major steps towards replacing people throughout the workforce all up and down the employment ladder. Now we have a problem.
Up until now, it has only been our ability to make sense of our surroundings and operate instinctively with hardly a conscious thought that separates us from automation, but that gap is starting to close, FAST.