Watson IBM computer wins Jeopardy

This week an IBM computer comprehensively defeated two former Jeopardy game show champions. Since Jeopardy is essentially a general knowledge competition, one might think that this feat isn’t particular impressive. After all, computers are able to reliably store millions of facts via databases; all a computer needs to do is access the piece of information via its memory. The difficulty with Jeopardy is the cryptic way that the questions are asked.  It would be impossible to program the computer with all the relevant information. Rather, the computer needs to understand what information it needs to access and ‘see’ connections between one piece of information and another. If a computer was able to draw out information contained within larger bodies of text, for example, it might have a chance. However, understanding natural language isn’t that simple.

Whilst I was an undergraduate studying Philosophy and Linguistics, I took a module in computational linguistics whereby we spent a whole semester programming a computer with natural language. After twelve weeks of entering determiners, pronouns, nouns and verbs, the limit of the computer’s language was to be able to to come up with a novel sentence such as ‘The cat laughed’. And it was probably able to come up with a handful of these type of sentences. It was a slow, boring slog with limited end results. But what it made me realise was that Alan Turing’s test (whereby a computer could fool a human into thinking it was human) was a long way off. Even if we, as humans, aren’t always perfect with our grammar, we can understand the difference between ‘The cat sat on the dog’, ‘The dog sat on the cat’ and ‘The dog was sat on by the cat’. Yet this is an immensely complex task for a computer. But now it seems as if a computer is able to understand language in a similar way; it is able to interpret questions and access the correct results.

So what does this hold for the future of artificial intelligence? There are some, such as Raymond Kurzweil, who argue that in the next 30-40 years superhuman (artificial) intelligence will spell the end for the human race as we understand it. When electronic technology has developed at an exponential rate over the last 100 years, it won’t be long before computers are able to understand, manipulate and construct the world in a (super)human way. Kurzweil predicts that by 2045, the computing power of artificially created devices will exceed the brainpower of all humans that have ever lived. That is quite a sobering thought and one that will have profound effects on our own existence. That doesn’t mean that computers will attempt to destroy the human race or enslave us as portrayed in science fiction films such as ‘Terminator’. It is more likely that we will start to amalgamate this technology into our own bodies.

The age of the cyborg will finally (after many false proclamations) have arrived.


Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s