When will a search engine understand like a human?

The blank search box is often a starting point for our journeys into the web. We search when we think it will be the quickest way to get what we want, whether it is on a search engine like Google, or using the search feature on a site such as Facebook.

Web searches have been classified into three categories: Informational (the user is searching for some information, not necessarily from any specific website), Navigational (the user is trying to get to some specific website), and Transactional (the user is starting their journey towards completing some form of transaction online, which could be making a purchase, a download or interacting with some online service).

It is useful to understand these motivations behind searching if we are to improve the online search experience. It is also important to understand what aspects of our brains are in-use when we type those few words into the Google search box, and hit enter (or click ‘Google Search’ if you are really old school).

Our brains are highly specialised information-seeking machines, but of course for the majority of our time on this earth, there was no Google or Yahoo to go to. This means that when we conduct a search online, we are using certain functions of our minds that have evolved because they were successful at other kinds of information-seeking. And because for most our history we haven’t even had books, our main source of information has been us! So it makes sense that the model we have in our minds of how to find information is really based to a certain extent on the model for a conversation.

We are well adapted to finding out things we want to know by engaging in conversations with other humans. Now it becomes significant to realise that humans are pretty different from search engines — we are highly intelligent and empathetic, and we can understand a broad scope of contexts and other factors that help to add meaning to the simple questions other humans ask us.

Now that we understand this, we can start to ask how search engines are meeting the goal of responding as a human would to our search queries, and how they can improve in the future.

Search engines themselves have undergone a radical evolutionary process in their short history. From early beginnings of simply matching search query text to text on web pages, the top search engines are now far more adept at comprehending the intentions behind our searches.

If you search for ‘Cape Town’, you are likely to be returned results related to current weather, recent news stories, and accommodation and holiday advertising among the usual search results. More and more we are being served with real time results from sources such as Twitter — perhaps a further move towards a more immediate conversational response.

However, if we were to ask a person about Cape Town, the person would know that what we are really interested in is some kind of specific information. The conversation would continue with the other person asking us if we want to know about the weather, or the history of Cape Town, or buying property there. In the same way a modern search engine throws out these options for you as part of the conversation of search, and you as the user take the next step by choosing from the options, a person would be able to assess whether our query was transactional, informational, or navigational. Further to this, they would often be able to understand the motivation and context behind our query, and because of that, give us a better and more accurate answer.

Search engines such as Wolfram Alpha and Powerset are striving to understand the content of a users’ searches so as to return more meaningful results, and this trend is clearly the focus of all improvements to the major search engines: Google, Yahoo and Bing.

The founders of Google once said that they wanted to document and index every piece of human knowledge. They also said that one of their ultimate goals is to create an artificial intelligence capable of interacting with a user just like another human.

Add these two together, and you get a new breed of search engine – one perfectly tailored to the task of providing us with the kind of search results we have evolved to expect.