Google has stated it is moving from a mobile world to an “AI first world.” So what does this mean for us and should we prepare to throw away our iPhones and Android devices?
In last week’s newsletter I discussed Facebook’s goal to create “intelligent machines.” Unsurprisingly, that goal is shared by Google (and its holding company, Alphabet). The mission of the Google Brain division, for example, is to “make intelligent machines, and to use them to improve people’s lives.” But whereas Facebook has just two artificial intelligence units, one for research and the other product-focused, Google has many different AI projects. Perhaps even one for each letter in the alphabet. Why is this? Because Google’s ambition for AI is much larger than Facebook’s.
As far back as 2002, Google co-founder and Alphabet CEO Larry Page is reported to have said “we’re really making an AI.” With the benefit of hindsight, the rest of us can finally see what Page was on about. Google search, still the company’s core product, has evolved into an Artificial Intelligence machine. It’s certainly the smartest thing I have daily contact with. A few nights ago, for example, someone on Twitter remarked that HAL (the supercomputer from 2001: A Space Odyssey) is one letter away in the alphabet from IBM. Or put another way: IBM minus one. That had never occurred to me before, so I looked it up on Google search. Immediately I saw that this mini-conspiracy had been discussed many times in the past on the Internet. Ok that’s a relatively insignificant factoid, but it would’ve been almost impossible to research pre-Google.
The Big Brain Theory
Google search may be the cornerstone of Google’s AI prowess, but over the past few years it has doubled down on AI technology. For a start, Google has acquired more AI companies than any other bigco – nine of them, according to CB Insights. The most significant was DeepMind, a British company that develops self-learning algorithms. You may not have heard of the company before, but you’ve probably heard that its technology defeated a champion Go player. Go is a board game considered much tougher for computers to master than chess, so this was a big step forward for AI.
DeepMind has larger ambitions than just playing board games. Its co-founder and CEO, Demis Hassabis, wants to build the first “general-purpose learning machine.” In other words, the goal is to build a computer that thinks like a human brain. There’s a computing term for this: artificial general intelligence (AGI). It’s also known as “strong AI” and has been the holy grail for AI researchers for many decades.
This strikes me as the biggest difference between Google’s goal for AI and Facebook’s. In both its research division and its product team, Facebook’s AI efforts are focused on automating tasks (for example, sorting photos) and augmenting Facebook users (via its prototype Messenger assistant, M). In other words, Facebook has specific user-centric goals for AI. Google, on the other hand, appears to be trying to replicate the human brain. Certainly that will improve Google search. But there’s more to Google’s AI ambitions than that. I think Google is trying to make AI into a utility, much like electricity or water.
Regardless of its long-term goal to make a giant AI brain, Google’s short-term goal is the same as every every other Internet bigco: build a voice-controlled assistant. Over the past couple of weeks, I’ve covered the leading contenders in this product category: Apple’s Siri, Microsoft’s Cortana, Amazon’s Alexa, and the indie company Viv (created by Siri’s founders after they left Apple). Google’s initial effort was an Android assistant called Google Now. However, at its annual developer conference in May, Google announced a new product called Google Assistant.
Google’s announcement blog post described Google Assistant as “an ambient experience that will work seamlessly across devices and contexts.” Of course it will be “conversational,” the trend de jour of 2016. But Assistant will be more conversational than Now; Google says it will engage in two-way dialogue with the user. Or as Google CEO Sundar Pichai put it, “we think of this as building each user their own, individual Google.”
Assistant will soon be deployed in three new, upcoming products: Google Home (a device that looks and functions like the Amazon Echo), Allo (a new messaging app which sounds very similar to Facebook’s M), and Duo (“a companion app for one-to-one video calling,” similar perhaps to Apple’s FaceTime). As usual, Google’s product naming is terrible. But you can see where it’s going with Assistant – this will be like a Google search box you can converse with anywhere and anytime.
Moving From Mobile First to AI First
Google signaled the importance of AI to its vision in this year’s Founders’ Letter, an annual letter to its stockholders. Normally the Founders’ Letter is written by Larry and Sergey, but this year it was penned by new Google CEO Sundar Pichai. He wrote:
Looking to the future, the next big step will be for the very concept of the “device” to fade away. Over time, the computer itself—whatever its form factor—will be an intelligent assistant helping you through your day. We will move from mobile first to an AI first world.
Google Assistant is clearly a pointer to this device-free future. But Google is a highly experimental company and so it has many other AI products on the go. A number of them are collected under the trendy term “machine learning.” The next big breakthrough in building the Giant Brain is just as likely to come from a small internal Google project, as it is from the teams behind Assistant or DeepMind.
The heart of Google has always been its search engine, arguably the world’s most sophisticated cloud-based software ever since its debut in 1997. Of course back in the late nineties, we only had PCs and web browsers to access Google with. Now we have mobile phones too. Soon, we’ll access Google’s Giant Brain anywhere and anytime, simply by talking out loud. That’s the “AI first world” Google is working towards.