Blurb: This article describes the different approaches experts are taking in the development of language in AIs.

Historians tell us our strange longing to confront mechanized mirrors of ourselves is nothing new. The Greeks had mused over how artificial beings could be built, and the ancient Egyptians, we are told, built religious statues that could — through clever riggings, pulleys, and secret compartments — talk and move their arms.  Nowadays,of course, we’ve set our expectations a little higher.  Thanks to science fiction, the term artificial intelligence (AI) is today likely to conjure imaginings of human analogs capable of — among other things — profound insight and rational conversation.  

The reality, however,  is that — in spite spectacular recent advances in robotics and AI technology such as Tersasem Movement’s Bina48  and IBM’s Watson — we are still a long way from seeing robots that are capable of sustaining what we understand to be intelligent conversation. 

Why this is so has to to do with the impossibly huge task of manufacturing machines that function like the human brain. 

Making electricity talk

The writer, James Gleick, describes how Isaac Newton retrieved old words from the English vocabulary and gave them extended — if not entirely new — meaning.  The words force, mass and motion had been around long before Newton, of course, but they held little scientific significance until he assigned them physical properties that stood up to the demands of mathematics.  For purposes of science, these ancient words had to be afforded quantifiable measures.  An analogous principle is constantly at work in modern computing.  When you send an fax through an office telephone (read more), the information you send is first reduced into bits of data that can be transmitted through the internet.  That data is then reconstituted into intelligible form on the recipient’s fax machine. 

In attempting to create artificial intelligence that talks and behaves like human beings, the complexity of the process is multiplied exponentially.  AI developers who want to build machines that replicate human behaviour and language confront not only the challenge of reducing words into bits of data, they also need to devise a system that automatically calculates the myriad rules that inform our social norms,  the processes that drive human learning, and all the other subtle influences that shape human consciousness.  This can by no means be an easy task. 

Schools of thought

In fact, in order to understand just how difficult it is to program machines that process and communicate information like the human mind , one need look no further than the heated, on-going debate among experts as to how the task should be accomplished in the first place. 

There are those in the field who agree with the noted linguist, Noam Chomsky, who has sharply criticized purely statistical methods in any attempt to mimic human thought, and desires an elegant theory of intelligence and language.  Then there are those who espouse the views of Google’s Peter Norvig, who argues that — by priming enough statistical data into a necessarily complex computational system– developers might one day be able to create artificial intelligence that can use language like human beings.

What makes us unique

Patrick Winston, who was director of  MIT’s Artificial Intelligence Laboratory from 1972 to 1997, offers still another take on the subject.  Winston believes developers should focus on things that make human beings unique. 

Two years ago, he told Technology Review that the magic ingredient lies in the human ability to create and understand stories using the faculties that support language.  “Once you have stories, ” he said, “you have the kind of creativity that makes the species different from any other.”

Indeed, poets and writers have always hinted that there is something magical about the process involved in transforming ideas into words, and words into communicable intelligence.  “The mind was dreaming, ” wrote Borges. “The world was its dream.”

Whether or not we will ever see machines that are able to reason and converse like human beings remains to be seen.  Clearly, it’s a tall order.  But Dr. Winston nonetheless believes we are “only one or two breakthroughs away.” 

If so, then maybe one day we can ask our thinking, reasoning robots to entertain us with tales about why they took so long to arrive.    

Monique Jones is an Engineer who deals with telephone systems. Monique graduated as a Cum Laude with a Degree in Civil and Communications Engineering. Besides being an Engineer, she also works as a part time Writer. She helps her colleagues and other people about their communication issues, giving effective solutions to address their needs. On her free time, she works on her fashion business, read books, and chat with friends. She also loves traveling and photography.