Upgrading the way we search
Think how you were browsing the web 10 years ago and how you are browsing now. Pretty much everything has changed, from the devise to the website you browse through. However, one of our primarily online usages has barely evolved over the past few years and that is the way we search. For sure the quality of the outputs are constantly improving, partly thanks to AI, but the way we are dealing with inputs is still very much the same. We type from a keyboard into a search box, while hoping to get lucky.
This stone age way of searching does have some obvious limitations:
- It creates an entry obstacle for countries where significant percentages of populations are illiterate.
- Even in developing countries, it happens to me all the time where I want to browse for a product or a person but that I don’t have the exact name. In that random search configuration, typing is not the handiest way to search.
In my view, speech and images are a much more natural way to communicate.
Speech search is ramping up quickly thanks to the massive improvements made by Siri, Cortana and Google now. The “Hey Siri” feature deployed with the iPhone 6 is in that way particularly impressive. But often these searches are limited to very general questions as speech implies that the user is not necessarily looking at the screen and therefore is expecting an oral answer. The usage of speech for search is no longer limited by the input but rather by the shape of the output.
In that sense, not only applying image recognition technologies for search will enable much more precise searches but it also implies that the user is holding or looking at a devise. Imaging willing to describe a handbag without knowing the exact name. It would result in rephrasing a typed query three or four times before finding out the name we were looking for. Given the recent progress in AI, pointing a smartphone camera could definitely be much faster. Think of Shazam for objects and faces. About this last aspect, in 2012, Google took the decision to avoid implementing facial recognition technology into its own search engine. Chairman Eric Schmidt said that he is concerned about potential misuses of the technology and predicted that some company would “cross the line.”
Apparently Baidu is moving fast towards such technologies and consumer demand is already there. “In five years, we think 50 percent of queries will be on speech or images,” Andrew Ng, Baidu’s chief scientist and the head of Baidu Research, said during a Gigaom meetup. “I think that whoever wins AI will win the Internet” ended Ng.