by I-COM | 04 July 2018
“The destiny of [Google’s search engine] is to become that Star Trek computer, and that’s what we are building.” - Amit Singhal, Google
In 2013, Slate published an article entitled “Where no search engine has gone before” all about Google’s obsession with building the Star Trek computer. It’s an interesting read, talking about the idea of having a conversation with a machine - speaking, not typing - and having it give you the right answer every time. In fact, the computer might even anticipate what you need.
It’s not just Google, though - Alexa, the assistant built into Amazon’s Echo device will tell you “I want to be the computer from Star Trek!” if you ask it “What do you want to be when you grow up?” and, as Ars Technica explains, the iPad and the Apple Watch very closely resemble handheld devices used across the different Star Trek series and films. The people designing our digital future are Trekkies, bought into not only its hopefulness about the future of humanity, but also the way in which its technology operates.
For anyone who’s never watched Star Trek (surely this must be a small number, even amongst non-Trekkies?), the crewmen and women on the U.S.S. Enterprise can operate their ship’s computer either by voice or via a screen. The computer can answer questions, control systems and ship’s functions (lights, temperature, engines, etc.), access data and carry on a conversation even as a person moves from room to room. The computer is also accessible via a portable, handheld device similar in size and shape to smartphone.
When you think about the way we search now, and the array of devices controlled by voice and by a basic AI, technology (and Google in particular) has come a long way towards giving us that experience of the Star Trek computer. Google has an assistant that you can address vocally. It can thread conversations (to a point), meaning you can ask follow up questions without having to re-use the trigger word (OK, Google), and they recently published some research into how they are developing technology that can better understand the intent of searchers’ queries so that they can produce more accurate results. This will be done by analysing the answers and comparing the different questions that may seem different but produce similar responses, as opposed to questions that sound the same, but need different answers.
The most recent statistics suggest that 40% of adults use voice search regularly - whether you’re talking to Siri on your iPhone, Cortana on your Surface, Alexa through your Echo or the Google Home Assistant on your Android device or through a Google Home speaker. Certainly, being able to speak to a fairly intelligent computer has come a long way.
When we’re not speaking into our devices to find the nearest coffee shop, the most convenient movie times or speaking to a customer service chatbot on Facebook, we’re queuing up to gain access to Amy, the AI personal assistant built by x.ai to schedule meetings for us busy office workers. Using this technology you can CC Amy into your email chain and she will take over the correspondence to set up appointments for you.
And then there’s Rankbrain, the AI which is reordering search results for Google. Even the team behind the search engine can’t explain why Rankbrain makes the decisions it makes. They’ve taught it what a “good” SERP (search engine results page) looks like and now it makes its own choices, based on what it thinks you want.
If the end goal of the companies driving our digital future is to create an AI that provides the experience of using the Star Trek computer, then they’re getting ever closer.
The voice assistants of today don’t always have the perfect answer - they get their results from search engines or other human-built resources - but that is why Google has been building its Knowledge Graph for years, in order to understand entities (a person, a place, a brand or even a concept or the characteristics of any of these things - any sort of thing that we can describe and discuss), the relationships between entities and to classify facts. However, with 16% of the queries typed (or spoken) into Google each day being completely new, Google will need to get far better at understanding context as well as at answering difficult questions, such as when someone is posing a philosophical or situational question, rather than asking for a fact. For this, Google will need a far-improved AI.
Tim Urban, from the superb Wait But Why blog, writes about a third party survey of hundreds of scientists that returned a result giving 2040 as the median year when we might see an AI that is as intelligent as a human. To reach superhuman intelligence? It could be hours after that, or years (and this is a fascinating subject in itself - the 2-part series on Wait But Why is worth the hour or two it takes to read, and it will delight and terrify you in equal measure).
Assuming what Google, or Amazon or Apple, really needs in order to build a Star Trek computer is an AI capable of human-levels of intelligence with access to a large search index - and the ability to interpret the context of the questions and critically assess all the information to which it has access in order to deliver a coherent answer - then we’re perhaps not that far off.
In the meantime, simply being able to adjust our heating while out of the house, or ask our phone for directions while we’re driving is futuristic enough to be getting on with.
This guest article was written by Mindy Gofton, Head of Marketing Strategy & Innovation at I-COM, a full-service digital agency based in Manchester.