Image by Mike MacKenzie on Flickr

In 1972 I was living in the Netherlands, near to the NATO military base on which my father worked. One Saturday my family spent the day at a fair held by various members of the forces to raise money for charity. There were colourful tents, marquees and enclosures spread out across a large playing field, where you could buy food and souvenirs, ride on a pony or try your luck at a game of chance. It was simple fun for all the family.

One tent in particular caught my attention, and I stopped with my Dad to have a look. There was a large kiosk in the centre of the tent called ‘The Knowledge Machine’ with a notice on the outside that read: “Ask me any question, and I will answer.” I wanted to try this, so my father paid a few coins and I took a sheet of paper from the table and wrote my question. I wrote: ‘Who was the oldest man in the world?’ folded the paper and slipped it into the slot. A few seconds later a note appeared from the slot. On it was typed: John Mosely Turner, 15 June 1856 – 21 March 1968, 111 years, 280 days. At the time, he was the oldest man to have lived, so the answer was correct.

I was still wondering how this had been achieved (there were no accessible computers in 1972 and the Web was almost 20 years in the future) when my father wrote another note saying ‘Wrong answer – the oldest man ever to have lived was Methuselah, who in the Bible lived to be 969 years old.’ He folded the note and inserted it into the slot.

There was another pause, and then a typed reply emerged. It read: ‘Methuselah falsified his birth records to dodge the draft’. We both laughed at the humour. Clearly, what was inside the kiosk wasn’t a machine, but a couple of guys (probably US servicemen from the style of response) having a lot of fun as they sifted through dictionaries, encyclopaedias and other tomes of knowledge to respond to any questions the public posted through the slot.

Recalling that story from my youth led me to think about John Searle and his Chinese Room thought experiment. In it, Searle postulates that a man is locked inside a kiosk, and there is a slot in the kiosk through which questions are posted, in Chinese. The man has no knowledge of how to read the symbols or understand the language. What he does have available are detailed instructions about how to process the symbols and respond with other Chinese symbols. His task is to take each question and respond appropriately in Chinese.

Those outside believe that a native Chinese speaker is inside the kiosk. In effect they believe that the kiosk contains a ‘mind’ that can comprehensively process morphographic, syntactic and semantic – and perhaps even cultural elements – of Chinese. What they don’t realise is that the process involves no knowledge of Chinese at all – just an operator systematically following a list of instructions to achieve convincing results. This is also the basic principle of the Turing test – are you talking to a computer or to another human? Ultimately, the test determines whether a computer can be considered ‘intelligent’. The Turing test is the measure of the extent to which machine intelligence can mimic human intelligence through complex pattern matching.

We’ve come a long way since 1972. Today, search engines and similar technologies can achieve results instantaneously, whether it is answering simple knowledge-based questions about the oldest man in the world, or more complex problems such as translating one language accurately into another. What’s more they are delivered directly to our devices, wherever we are, at any time of the day or night. All we need to do is ask. The answers seem intelligent, and we are comfortable conversing with technology to the point that we entrust it with our daily quest for knowledge. A lot of artificial intelligence is built into the search algorithms and natural language processing it employs. And yet we are still a long way from the point where we can converse with technology as we would another human. If we could, would it be desirable? I wonder how far down the artificial intelligence road we have travelled to the point we see a computer spontaneously responding with humour?

The knowledge machine by Steve Wheeler was written in Plymouth, England and is licensed under a Creative Commons Attribution-NonCommercial-ShareAlike 3.0 Unported License.