A newly born child begins from its very first day to make conclusions about this strange place it has been born into, and to store them away. Over time, this tiny collection of data and conclusions becomes an immense library of knowledge. That – the ability to develop such a library and to act on it – is what is normally meant by “human intelligence.”
The question is: if a computer is taught – that is, programmed – to analyze data and draw inferences, does this equate to human intelligence or is it simply the exercise of a learned operation?
The reality is that some will say, yes, that is intelligence and some will say, no, that is not enough of a cognitive function to represent intelligence. That argument is not going away anytime soon; some will believe they see artificial intelligence and some will believe they do not.
Nevertheless, a great deal of that kind of work is being done. Digital Reasoning is a cognitive computing company that has been around for 16 years now and has worked with the American Department of Defense to seek out and identify terrorists and with American law enforcement agencies to find examples of child sex trafficking in America.
Part of what Digital Reasoning’s software does is straightforward: it analyses online advertisements and conversations to find overt cases, particularly of sex trafficking. If it were limited to that, it would be inadequate for the job in hand because most people involved either in terrorism or in sex trafficking do not express openly what they are up to. The software, therefore, uses machine-learning technology to analyze language and to draw attention to changes in behavior – someone talking less than they did, or more than they did, or in a different way, or setting up new network links with others already under suspicion. It is the system’s ability to learn over time, by analyzing the data it goes through, that causes its proponents to say that this genuinely is artificial intelligence.
For most observers, the jury is still out.