AI: Where is the Human Touch | Mirum India
(This article was published in Businessworld)
Driverless Cars, Chatbots, Artificial Intelligence: where is the human factor?
“The machines will take over!”
That has often been the fear, even several decades back, when the first hints of automation started to be seen. So far those fears have been unfounded, and humans have still remained largely (or so we believe!), in control.
It’s another thing that you see kids (and adults) glued to their phones 18 hours a day, and you wonder if that machine / device has indeed taken over life!
But that is an aside.
The real fears with technology is about the human being, being replaced! And while simple automation including robot run factories, did not quite achieve that, new fears abound with the emergence of artificial intelligence, and with services like driverless cars and chatbots, for example.
And then there is the other aspect – even if humans don’t get replaced, with AI taking over, will we lose the “human touch”? Or will all interactions become very mechanical and emotion-less??
First, let us consider the possibility of humans being replaced.
When earlier aspects of automation came, the most famous being large automated factories, those did take away many jobs. And yet, in the larger scheme of things, the factory jobs represented a small fraction of the working population. And also, for whatever reason – maybe the cost, maybe the real utility, maybe the need for extremely well-defined process – these robots did not really take over ALL factories. Except for assembly lines like automobile factories, or some hazardous chemical / nuclear establishments, where in fact, it was a good thing to replace humans, the rest of the factories still continued to have humans running them.
But now comes the era of Artificial Intelligence. And if early use cases and demos that we have seen are any indication, this technology threatens to be far more pervasive than robots and mechanical automation ever was.
The use cases of AI that we already see in action today include:
- Driverless cars: that can drive safely, understand road hazards, understand speeds and traffic lights, and pretty much navigate the vehicle, like the best of trained human drivers can do
- Chatbots: that respond to all those personalised queries that people have, when they reach a website or an app or on a messenger service, and which are understood, clearly analysed, and responded to extremely correctly, sometimes better than what a good trained human might do, in their place
These are a couple of simple examples which are gradually getting to be seen, in practise, beyond alpha trials.
But there are many others that have been demonstrated, for example, at the IBM Watson showcase events. Where AI has shown the creation of a song, or a painting, where the “machine” has completed a complicated financial audit exercise in a matter of a few minutes, what a typical Big 4 audit firm, would have taken several weeks to complete, etc.
Clearly the ability of AI to “learn” a process, and then rapidly improve to the point of being precise, and the ability for programming to enable that learning, “threatens” pretty much, all kinds of repeatable tasks. And let’s be clear – most human beings involved in “work” today, are doing repeatable tasks. Whether making and service ice cream cones at a fair, whether mowing a lawn, whether processing insurance claims, or cleaning the glass façade of your building. So the fear is whether an extremely large number of human jobs will get threatened, thanks to AI?
And where the one view heard often is that inflexion of this kind happened in the past as well, with technology displacing some jobs, but humans retrained themselves to do other things, and typically moved up the value chain of work.
The fear this time, though, is that AI threatens to disrupt an extremely large portion of work, across domains, all at the same time. Never before have we seen, for example, 60% of all workforce getting threatened for possible displacement, through a technology development, all at the same time! And for the world economies to provide alternate options for such a large number of people!
It is perhaps for such fears, that stalwarts like Bill Gates and Elon Musk, amongst others, have expressed their concerns over the increasing penetration of AI technologies.
These fears notwithstanding, technology once available, is difficult or perhaps impossible, to be reined in. Besides other reasons, the most important driver is almost always, a commercial one. If a business for example, can replace 100 human chat operators, with the implementation of a successful chat bot service, and save that kind of money, they will do so. Or if a truck operator, operating a fleet of a 1000 vehicles, can replace drivers with automated trucks, and save a bunch of money on ongoing expenses, they will do so too.
So you cannot turn back technology, and AI will run a lot of work tasks, in the near future. What happens to the people whose jobs get replaced, is a different question, and hopefully, mankind will find answers to that.
But do we also fear the AI to give us very synthetic, almost artificial experiences, rather than the human touch that we normally see, for example, when we chat with a person, or when an Uber driver, drives us and tells us local stories??
Well, the fears of losing that touch, are not unfounded. And earlier experiences with any kind of automated service, have shown it to be very “robotic”! That is indeed, from where the adjective, robotic, emerged. However, this is also where AI is different. Besides “learning” the actual process – say, of driving a car – AI can potentially, also learn to “be like a human”.
For example, the driverless car, driving the passenger just landed at an airport, to his hotel, could “ask” the passenger about his flight, where he is coming from, whether he finds the AC in the car comfortable, whether he would like to listen to some music, etc. And the same car could talk like a young girl, or a mature man, or whatever kind of response one desires to make happen.
So the human touch need not entirely be lost. Of course, all of these need “training”. For the AI engine. And an investment into that area, would ensure that the human touch remains.
But as you can imagine, this training will be for “good and predictable” behaviour to begin with. If for you, the real human touch is an occasional chat operator who bumbles away and makes a mistake, or a driver who may smell or may get abusive once in a while, and if you miss those “real” human experiences, chances are that those kinds of human touch, may not be seen easily, in an AI scenario! Not that those CANNOT be programmed too, but I don’t think those are going to be high on the priority of AI trainers!
Sanjay Mehta
Jt.CEO, Mirum Digital Pvt Ltd.