Google has recently acquired a major AI (Artificial Intelligence) developer called Deep Mind in the UK. This is the largest Google acquisition to date in Europe.
Apparently Deep Mind already worked on search before the acquisition.
Their patents certainly suggest that. Deep Mind is just the latest step on a long path for AI-obsessed Google though. Where will it lead?
What is AI?
To understand what AI means you ideally should watch the movie “I, Robot” from 2004 starring Will Smith and filmed based on a novel or rather a set of short stories by sci-fi grandmaster Isaac Asimov.
First off Artificial Intelligence doesn’t mean necessarily robots like many sci-fi movies suggest.
The movie that often comes to mind first when thinking about Google AI – the Terminator series from the nineties – does combine AI and robots (or cyborgs) in the plot so much that you end up believing it’s the same thing. That’s why I recommended “I, Robot”. In it you clearly see the difference between machines and artificial intelligence. The robots are very advanced in it but still just obeying orders from humans and acting according to pre-programmed routines. Initially at least.
Real artificial intelligence is effectively thinking for itself. It can go beyond the built-in programming. It can evolve. AI can learn for itself.
Isn’t AI just science fiction?
This is also the major part of how AI gets seen by the general public. It’s just scary science fiction, isn’t it? No, it isn’t. AI is not self-aware yet as the Skynet creator asserts but the idea itself seems to be feasible by now.
We also witness first steps into that direction: Google has already been involved in some of them. Quantum computers for example bring us to the verge of AI. Learning algorithms are already being built by Google.
What can AI do for Google?
AI can predict what you want before you want it for example. That’s indeed one of the closest developments we can expect from Google AI. They are working on methods to find out what you will search for before you actually start typing or saying it. It can determine intent based on the context.
Who you are, where you are, what time it is can be combined to predict your needs.
In a way it’s not “rocket science”. When I get up at night at home and search the kitchen drawers it’s logical that I might need sleeping pills for example. I could of course search for something to eat too but combine that with your smart fridge that knows what to buy itself and you get the point. So you’re presented with Google assistant who also can scan your home (another acquisition by Google) and will tell you where the pills or the food are.
How will AI affect the search experience?
By now Google wants to be your personal assistant. The standalone service is called Google Now but it’s already integrated with overall search results. So when you search for “weather” you don’t get 10 links to weather sites but just one full page result from Google itself, the weather forecast for your area. Google automatically detects where you’re at. This is not AI yet.
The Google assistant will one day be powered by AI.
Google could check your calendar and find out that you don’t work today. Then it would check the weather for the area and your latest purchases, like say a tennis racket and then it would suggest the best nearby places to play tennis for beginners even before you plan the day. The smarter the AI the better it can predict your needs or provide you answers for questions you haven’t even asked yet.
AI would be the perfect servant. You wouldn’t even have to pay for it.
Of course the owners of the “websites” or rather services would. Soon enough Google will merge paid and organic results completely in one way or the other and you will have to spend money taking part in the Google Now network just like with Google Shopping.
How to optimize for artificial intelligence?
I’m afraid it’s too early to say but I can guess some of the things that will help. Right now on the open web hyperlinks are still determining ranking factor on Google. We have seen the trend of Google trying to curb the use of links “for SEO purposes” on the other hand.
Google attempts to replace links with their proprietary Google+ features like +1 votes, authorship identification, Google+ shares, comment and post embeds. AI will be able to understand the whole context of you and your relationships based on those and other identifiable signals tied to you as an individual.
A search algorithm is still a very primitive computer program until now.
At the end of the day it just counts links and their value, and adds or subtracts numerous other less important signals. Until recently it also attempted to count words to determine the relevance of a text.
With newer semantic technologies the algorithm makes the first steps to “understanding” what you are searching for. With AI this wouldn’t be a static process where a fixed algo is trying to determine the connections following pre-written programming.
AI could react on its own to find out what’s going on and provide solutions independently by itself.
Right now when there are only 10 options provided the algo can only choose one. In future the AI could come up with an eleventh option based on the context. So to optimize for AI you have to provide the right type of context.
Should we be scared of Google AI?
My grandmother was born in 1899. She once told me the story of how the first cars appeared in her area. People on the streets were scared she told me. They assumed cars without horses were run by the devil. It’s not the devil you have to fear we have learned by now. It’s other humans and how they use inventions.
Of course motorized vehicles are a problem by now to some extent. Just think of tanks and such. Today cars are so permeating our culture that we tend to live for our cars instead of using them. The air is polluted , the streets congested, kids get run over and thousands of people die every year in car accidents. Similarly,
it’s crucial to anticipate imminent problems stemming from new technologies.
How will AI be used? What for? Who will control it? Google is not a charitable organization and its power is already overwhelming. So handling the development of AI over to Google might be a mistake. Of course profit motivated development might be faster than one based on democratic decisions by the people. The direction might be questionable though. We always have a choice on a personal level and also as a society. Do we want a place for our children dominated by cars or AI? I’d rather ride a bike and think for myself.
Google, robotics and AI Resources
I could barely scratch the surface of the AI topic so I compiled a small list of resources for you to read more.
- Conversations With Google
- If this doesn’t terrify you… Google’s computers OUTWIT their humans
- Google’s future: microphones in the ceiling and microchips in your head
- More on DeepMind: AI Startup to Work Directly With Google’s Search Team
- Artificial Intelligence and Machine Learning – Research at Google
- Google, AI and the future of SEO
- Google buys AI firm DeepMind to boost image search
- With Boston Dynamics, is Google turning into Skynet?
- The Day Google Became Skynet
- Is Google Terminator’s SkyNet?
- Google Puts Money on Robots, Using the Man Behind Android
- Skynet is born: Google making robots
From the Motion Picture I, Robot feat. Will Smith
