When will the Robots take over the world????
There was an eruption of euphoria about 1.5-2 years ago because of a phone call made during the unveiling of a slew of new tools by Google - A telephone call to book a table for dinner at a restaurant which was made by software mimicking a human being so completely that while listening to the conversation between the machine and the person who picked up the phone at the restaurant, you could not tell which was a person and which was a machine.
This was without question a spectacular feat of technology. And this seemed to reinforce the current, commonly accepted vision that one by one, the abilities and techniques that humans once imagined were uniquely their own, have begun to be taken over by ever more powerful machines and that in not too distant a future machine intelligence will surpass EVERY aspect of human intelligence.
People like Nick Bostrom have written influential books (Superintelligence: Paths, Dangers and Strategies) forecasting a dismal future of potential slavery or destruction of human beings by superintelligent machines.
However here is an interesting essay by David Deutsch (One of the pioneers of Quantum computing) - raising some fundamental questions about this popular vision.
https://aeon.co/essays/how-close-are-we-to-creating-artificial-intelligence
Deutch is particularly scathing in his assessment of what the field of AI (or AGI as he calls it) has actually achieved.
I quote from his essay -----------------
"The enterprise of achieving it <i.e. Intelligence> artificially — the field of ‘artificial general intelligence’ or AGI — has made no progress whatever during the entire six decades of its existence.
I cannot think of any other significant field of knowledge in which the prevailing wisdom, not only in society at large but also among experts, is so beset with entrenched, overlapping, fundamental errors. Yet it has also been one of the most self-confident fields in prophesying that it will soon achieve the ultimate breakthrough." ---------------------
Interestingly he firmly believes that Artificial General Intelligence as complete as human intelligence can be created - and is in fact warranted by laws of physics!
And I quote again -----------------
"AGI must be possible. And that is because of a deep property of the laws of physics, namely the universality of computation. This entails that everything that the laws of physics require a physical object to do can, in principle, be emulated in arbitrarily fine detail by some program on a general-purpose computer, provided it is given enough time and memory. -----------------
But he believes that the AI technology that is being pursued presently will never result in true general intelligence because of a fundamental assumption that seems to be being made about the nature of knowledge i.e. the belief that New knowledge can be created "purely" by "Learning" from past experiences, behaviors and outcomes.
and I quote again ------------------
What is needed is nothing less than a breakthrough in philosophy, a new epistemological theory that explains how brains create explanatory knowledge ------------------
The keyword here is "Explanatory" knowledge! A testable hypothesis that can explain WHY the output is what it is, that goes beyond using extrapolations from historical data which have been used to generate the output in the first place, as the only justification.
Deutch has also written a wonderful book called the "Beginning of Infinity" in which he treats this topic of "artificial intelligence" in much more detail. But if you don't have the time to read the book in full, Read this essay and then decide whether you agree with him or not.
We did some interesting thought experiments when analysing the capabilities of Amelia the AI agent of IPsoft. There were two fundamental principles which will limit AI 1. First was selfawareness of multiple planes for operating in a society.(Computational, Moral, Ethical, legal, sensory plane .....) .AI Watson or Amelia or which ever AIs function only in the computational plane. So if the doomsday scenario is possible (Hitler, Stalin type AI) the counter argument of an ethical AI (Like Mother Teresa or Albert Schweitzer) should also be possible. But both are impossible because not being selfaware limit the possbility to understand other dimensions. 2. The AI thrust was on what it can do and is a important corollary of self awareness (Not possible in IBM Watson Jeopardy, Google making humanlike phone calls or beating GO champions) . That corollary of self awareness is, that while knowing is important, most AIs will not know what they do not know or what they cant do. So Knowing what you cannot do is as important as knowing what you can do. An AI just does it without knowing and fails on both counts of knowing. So when David Deutsch speaks of "a new epistemological theory that explains how brains create explanatory knowledge", it already stares us in the face. Self Awareness or only Awareness. Self awareness on multiple planes. An ameoba or plant or a monkey or an elephant or a dolphin and a human represent awareness on different planes. Some planes are common for all living beings, self preservations, and some specific to an operating area, while elephants can use Ultrasound to communicate, humans cannot. The AI has to be rephrased as Computational power in a similar manner to the horsepower and speed specifications of a car. Once we remove the AI tag to computational power, we will see IBM Watson, Googles GO computational engine in a different light. So IBM Watson would be Computational power to listen to the question and quickly look up answers to that questions with probability algorithms and a predefined context framework for the question. And google GO challenger would be computational power to run algorithms to understand 1 million possible moves and find the best move out of the 1 million combinations AI is a hyped up mirage we seem to be following by pointing to a possibility of creating AI without self awareness.