The recent spat between Elon Musk, the Chief Executive of SpaceX and Tesla, and Facebook’s Mark Zuckerberg has added fresh fuel to the public discourse on the future of Artificial Intelligence (AI), which is at the heart of Facebook. It thrives on data which AI knows how to handle and analyse best. It is not concerned over privacy.

However, Musk believes that AI can take us nearer to the apocalypse, by running amok. The debate is decades old, but the future of AI may defy our wildest imagination.

When Garry Kasparov, a world chess champion, had lost his famous match against the supercomputer Deep Blue in 1997 after beating it the previous year, Time magazine had commented, “Luddites everywhere were on notice: here was a machine better than humankind’s best at a game that depended as much on gut instinct as sheer calculation.” A distraught Kasparov sighed and rubbed his face in disbelief before abruptly walking away.

That was the first time that a machine had beaten an expert in a game that requires intelligence. Since then, milestones have been breached at regular intervals.

In February 2011, history was made again, when IBM’s supercomputer Watson beat two contestants on a TV serial called Jeopardy, answering questions that they failed to answer and processing data at an astonishing speed of 500 GB per second, with a RAM of 16 terra bytes. It had access to 200 million pages of material in its memory which it could analyse on live TV within seconds.

Watson won the $1million prize money. We are living in the era of Artificial Intelligence in which machines have been taught to think like humans and act.

They still lack self-awareness, an essential attribute of human consciousness and cognition. Robots do not yet know that they are robots. But given the exponential growth of technology, it is only a matter of time before they acquire self-awareness that rivals human intelligence.

The logical conclusion of this process will be the development of machines whose artificial intelligence will surpass human intelligence itself, imbibing superintelligent machines to whom no problem will be difficult enough to solve.

Of course, the advent of such super-intelligent machines will take some years yet to happen. Human intelligence is essential for reasoning, planning, learning, communication (using the rules of natural language), perception, decisionmaking, and the ability to manipulate objects and perform complex tasks.

AI today can handle only some of these tasks, such as visual perception, speech recognition, decision-making, and translation between languages, writing reports based on data analytics, etc., though the AI wave-front is being expanded to include new areas of cognition almost on a daily basis. Today’s computers have super memories; they can easily perform millions of calculations every second and perform consistently at peak levels combining peak skills for an almost indefinite length of time. Using these calculations, they can break up an object into its tiniest elements ~ pixels, lines, circles, triangles, squares.

This is the way a machine tries to recognise patterns and identify objects. It can possibly identify a rose, but will have no idea about the ‘rosiness’ of roses. It cannot yet attribute any quality to an object, something the human mind does effortlessly. Machines do not also understand the simplest of things about our physical world, let alone the subtle nuances of the working of the human mind with all its complex web of emotions. Robots are just machines that can be programmed to perform certain cognitive tasks, but unlike humans, they cannot anticipate and plan for the future either, at least not yet.

But as human knowledge migrates to the Web, the horizon will expand exponentially. Then they will be able to read, synthesize and intelligently apply all human-machine information, and their non-biological intelligence will someday probably be able to match and then surpass ‘the subtlety and range of human intelligence’ in almost every sphere, as the futurist Ray Kurzweil visualises in his book, Singularity is Near.

But intelligence also needs emotion; without emotions, we would often find ourselves constrained to make judgments and decisions. The problem is, how do you create a robot with emotions and a value system? Emotions often defy logic and AI is dependent on logic and logic alone.

“Turing Test” is recognised universally as being the touchstone on which the intelligence of a machine is to be measured and determined. Turing had once said, “A computer would deserve to be called intelligent if it could deceive a human into believing that it was human.” This simply means that the responses of the machine to a given situation would be no different from that of a human being.

At Meiji University, Tokyo, scientists have taken the first step to create a robot with selfawareness by creating two robots, the first of which was programmed to execute certain motions, and the second to watch and mimic the same. The next step was achieved in 2012 when a robot was created at Yale University that passed the mirror test ~ the ability to recognize oneself in a mirror, which is considered a proof of having a concept of self.

Whether it would be an intelligent design or Darwinian evolution of robotic intelligence is a question which is still openended.

Just as our intelligence was not designed by an omnipotent creator but had evolved gradually, it is not necessary that we have to design the intelligence of machines we create and teach them how to think; in all probability they may not think like us, but they will ‘think’ nevertheless. As Quentin Hardy of the University of California at Berkeley says in his essay, The Beasts of AI Island, intelligence is merely a toolbox we use to reach a given goal, which doesn’t entail motives and goals by itself.

The new “Age of Thinking Machines” may even force us to fundamentally rethink and redesign our institutions of governance, allocation, and production which today are far from being perfect. Never before in the history of humanity have we experienced technology changing the entire landscape of the manmachine paradigm so swiftly and so profoundly. Revolutionary breakthroughs have been achieved in image recognition, data analysis, autonomous learning, and the construction of scalable systems.

These have spawned applications that were inconceivable only a decade ago, giving birth to systems that display significant language skills, skills for manipulating objects, learning and problem-solving abilities, factual and procedural knowledge and even some rudimentary imagination. AI is increasingly replacing human decision-making in many areas of cognizance ~ routine administration, engineering and construction, design, data analytics, and even robotics and AI programming itself.

Of course, there is still a long way to replicate human intelligence, but it may not lead to any humanlike intelligence. It may usher in an era in which there will probably be no such thing as ‘pure’ human intelligence, because all humans will be a combination of biological and non-biological systems which will constitute integral parts of our physical bodies, vastly expanding and extending their capabilities. Humans and machines will merge together to create a human-machine civilisation.

The initial impact will be highly disruptive and there will be ethical, socio-economic and other unsettling issues that will have to be addressed requiring a level of maturity that humanity has not yet perfected. As Quentin says, “We’re building new intelligent beings, but we’re building them within ourselves. It’s only artificial now because it’s new. As it becomes dominant, it will simply become intelligence.

The machines of AI Island are also what we fear may be ourselves within a few generations. And we hope those machine-driven people feel kinship with us, even down to our loneliness and distance from the world, which is also our wellspring of human creativity.”

Man and machine will then become one unified, hybrid entity. These entities will learn to think, emote and empathise in their own ways which might be different from our human thought processes, emotions and expressions of empathy. As the author Clifford Pickover writes, “We’ll share our thoughts and memories with them. We will become one. Our organs may fail and turn to dust, but our Elysian essences will survive.

Computers, or computer/human hybrids, will surpass humans in every area, from art to mathematics to music to sheer intellect. In the future, when our minds merge with artificial agents and also integrate various electronic prostheses, for each of our own real lives we will create multiple simulated lives.” 

(The writer is a commentator and the views expressed are personal)