Google conversational search is a bold leap forward

The first thought that crossed my mind when I tried out Google’s new “conversational search” functionality in the latest version of its desktop Chrome browser was Star Trek.

You don’t have to be a huge fan of the sci-fi TV series (and films) to remember the ways in which Captain Kirk (Picard, Janeway, etc) did search or asked a question. “Computer?” they’d start saying, and then speak.

That also reminds me of the rudimentary speech commands with the first-generation Kinect for the Xbox. You’d start voice-command interaction with that device by saying “Kinect?” and then tell the device what you wanted.

With Google’s conversational search, you don’t start with commands like those, you just ask a question at the Google search screen in your Chrome desktop browser.

This feature has been part of Google’s search functionality on mobile devices running Android and iOS for a while. The big difference, though, is using it on the desktop browser, the “computer speaks back to you.”

It’s actually very neat. Here’s a short video of a very quick and simple test I did, asking Google search, “How far is it from Wokingham to Hammersmith?”

Given the conversational way in which I asked the question, and my own accent, word-pronunciation and grammar usage, I was hugely impressed with the function’s ability to correctly – and very quickly – understand what I said. The voice answer was great, much better than the Max Headroom-style I might have expected.

My test is very simple. A clue to the real attraction of this new functionality for search lies in its description: conversational search. This is about asking a series of questions, each relating to the previous, like a developing conversation. I asked just one question. So what’s it like if you conduct a series, a conversation?

Danny Sullivan has a great detailed review of his experiences with doing precisely that, with impressive results.

Google conversational search isn’t perfect, far from it (and Google doesn’t claim it is). The new functionality had some teething troubles yesterday when it launched. And correctly recognizing every spoken¬† word is nigh on impossible – look at the trouble we humans have.

But it’s a huge step forward.