Blog: The User Experience Of Artificial Intelligence, Part 3
The many definitions of intelligence
There are many definitions of intelligence, and they reflect both the time in which they were proposed and the assumptions of those who introduced them.
In his book Intelligence Psychology, Christopher Robertson briefly synthesizes definitions, theories, measurement systems, and types of intelligence.
According to Robertson, the definitions proposed by groups or organizations are eighteen, while those proposed by psychologists are even thirty-five. Many of them are similar, but the number remains impressive. Moreover, Robertson lists several types of theories about human intelligence: psychometric, cognitive, cognitive-contextual, biological, and so on. And we cannot forget Jean Piaget’s theory of cognitive development.
So, who’s right?
If we want to call a machine “intelligent,” what kind of intelligence should it have? Environmental, musical, logical/mathematical, interpersonal, bodily-kinesthetic, linguistic, intrapersonal, spatial, existential-spiritual?
Regardless of the type of intelligence, how should artificial cognitive abilities develop? Gradually as in humans, thanks to the progressive development of the cognitive system, or are they supposed to be embedded in the machine from the beginning?
To implant a cognitive ability in a machine, the designer must know precisely how this ability works in the human brain. Otherwise, how might he design an algorithm to emulate it?
On the other hand, to build a machine that can learn in the same way we do, you must know exactly how the human learning process works.
If you think we deeply understand the human brain, that’s fine. We’ll have no problems reproducing it. Honestly, I don’t think so.
The problem of artificial intelligence is in his name
I do not dispute the fact that Turing was right and that sooner or later there will be non-organic machines full of NAND doors that can perform any computation on a human level and beyond (learning, thinking, creativity, and so on). We’ll see (probably not us). I criticize the attempt to anthropomorphize everything today for purposes that are useful only for marketing strategies.
The idea of artificial intelligence has undergone a remarkable evolution in the last sixty years. Starting from the hard cognitive science of the sixties, and going on until neural networks and more. However, it is essential to understand that today, cognitive science and artificial intelligence are distinct disciplines.
This fifty-year long history in which cognitive scientists enthusiastically associated the exponential growth of the computational ability of computers with the possibility of simulating, or overcoming, the human mind finally ended.
It is this long history of hype on calculators power that caused the improper use of the word intelligence. A word that can generate confusion, fear or even a sense of the ridiculous if associated with agents slightly more complicated than a thermostat, as well as the adjective artificial when related to elusive concepts such as emotion or consciousness.
The foundations of contemporary artificial intelligence are mathematics, economics, neuroscience, psychology, philosophy, computer engineering, automatic control, cybernetics, and linguistics — a constellation of competences and visions of the world that are remarkably different. Applications range from understanding natural language to handwriting recognition, automatic translation, robotics, collaborative filters, search engines, facial recognition, social network analysis, sentiment analysis, autonomous cars, and so on.
Today, the so-called intelligent agents are digital and physical devices that enhance the human experience of a service or a product and, likely, improve our quality of life. There is no consciousness, intelligence, or emotion in these artifacts.
Tomorrow, we shall see.