I’ve been thinking about how Google will one day be out-googled. Here I’m using the verb “to google” in its ancient (up to about 2002) sense to mean outdoing the competition, completely taking over the market. For example, “The new internet startup googled their established rivals.”

Google (or perhaps I should say “Google Pure”) is used in many ways, sometimes as a search engine to search for sources of information, but sometimes also as an answer engine to search for the information itself. Sometimes I want to find a page or a particular page that discusses medieval notions of obesity, but other times I simple want to know the average weight of men aged 30-50 during the middle ages.

An answer engine will have to trawl the internet, like Google. But instead of just indexing the pages, it has to perform a textual analysis of the information, extracting and classifying the facts (or at least the statements). Then, later, a user could submit queries such as “What is the fourth closest star to the Earth?”, “When did the French Revolution start?”, “Which is the hottest city with more than 250000 inhabitant in Brazil?”, or “Why did the Russians invade Afghanistan?”

Analyzing texts and answering such comprehension questions has long been a research activity in AI. I don’t now what they have been up to lately, but I remember that, when I was undergraduate, these were heavily researched topics.

And it does not have to end at text. In the future, I am sure, we’ll be able to ask the answer engine questions such as “Show me a picture of the Colloseum during May” or “Find a photo of Donald Knuth and Richard Clayderman from 1980-1990.”

I hope that someone somewhere is working on this.