During the Christmas period a friend asked me why I was stood in the queue at HMV rather than buying online. My answer, that I didn’t know what I wanted and needed to browse, made me realise that computers can only help me with the science of buying something, not the art of choosing what I want to buy.
When it’s a specific book I won’t think twice about heading to Amazon and buying that exact title, if I want to buy a relative some wine I know a site’s experience in the art of wine buying will make up for my lack of wine knowledge. However, I can’t ask an online bookstore to suggest £30 of good books or DVDs to me, “good” can’t be translated into a catalogue question that the web site will let me ask it.
Today I have to solve this problem by using a person skilled in an art, in this case selecting books. When I was looking for a hardback book, with between 100-200 pages, plenty of colour photos of Asia that someone aged between 40 and 45 would appreciate I got the answer by asking a store assistant’s knowledge of their store’s catalogue combined with her experience of people’s buying habits; using the art of book selection, not the science of catalogue querying.
Another example of this art vs. science challenge, this time topical, is America’s suggestion that it will use racial profiling to determine who to screen more thoroughly prior to boarding a US-bound aircraft. The assumption is that computer algorithms can analyse passenger lists and identify who statistically is likely to be a higher terrorist threat. The immediate flaw is that none of the recent successful terrorist suspects in the West fit a common profile, that’s why they were successful!
What we’ve seen in technology terms are two examples of the gap between in abilities between data mining and artificial intelligence. Data mining algorithms have been written which query a bookstore’s previous order data using scientific principles creating an output that says because you bought these 8 books you must like these further 8. These can produce results, regardless of how accurate, because consumers leave huge footprints behind them; a successful terrorist’s skill however is leaving none.
To conclude, today’s successful IT systems must recognise where science has to stop and an art must take over, science may never be able to replace art. If you’re building an e-commerce site plan for first time buyers who don’t know what they want from you, you’ll never replace an experienced sales person say “How may I help you….?”