Since 2009, $17 billion have been invested in artificial intelligence, reports the business analytics firm Quid. And in the past four years, investment has been growing at 60 percent a year. The best minds in Silicon Valley are betting on it. It’s a footrace to be the biggest, fastest. The law of
increasing returns to market leaders suggests that the more data is gathered by the leading company, the more likely that company will maintain its grip. Google is vying for that spot, and at the end of 2013 the company made a stunning announcement. In a six-month span, it had acquired eight of the leading robotics companies:
> Schaft, Inc., which won the Defense Advanced Research Projects Agency’s (DARPA) Robotics Challenge Trial with a two-legged machine that can climb ladders.
> Industrial Perception, Inc., which specializes in 3D vision for robots in factories.
> Redwood Robotics, which specializes in building robot arms.
> Meka Robotics, which builds “human safe, human scale” robots that can co-exist with people in homes and in the workplace.
> Holomini, which creates omnidirectional wheels for robots.
> Bot & Dolly, which specializes in motion-controlled cameras for film and television production.
> Boston Dynamics, which builds biomorphic robots such as the humanoids Atlas and PETMAN, as well as a menagerie of four- and six-legged creatures known as BigDog, Cheetah (and its successor, WildCat), and RHex.
> DeepMind Technologies, which builds artificial brains.
This spree was followed by several other acquisitions, including:
> Titan Aerospace, which makes unmanned high-altitude aircraft.
> Jetpac, Dark Blue Labs, and Vision Factory, all of which develop artificial intelligence software.
Just a few weeks later, the longtime AI proponent, scientist, and inventor Ray Kurzweil was named Google’s director of engineering. Some industry watchers were puzzled by this move. Why would a software company like Google be interested in tackling the messy hardware problems involved in making robots walk and roll and run? And this was only a short time after Google completed the process of dismantling and selling off the hardware divisions of Motorola, one of the last great American electronics companies. They miss the point. Just like Google’s entry into the automobile industry, this is not about hardware. It’s all about data.
EXTRACTING VALUE FROM SMART MACHINES
Soon, just like smart devices and smart cars, mobile robots will be connected to the Internet, and Google clearly intends to intercept and organize this data before it filters out onto the Web. Google’s robot strategy may be the biggest and boldest Big Data strategy of all.
In a way, Google’s strategy is a more audacious strategy than Microsoft’s in the mid-1990s. Back in its glory days, Microsoft inserted a wedge of proprietary Windows software between computers and the applications that ran them. Today Google is replicating that strategy across an astonishingly broad range of devices. Like Microsoft, Google seems to be less concerned about the actual
hardware (car, computer, phone, dishwasher, or robot). It will build these things if it must, especially
to make a reference model like Nexus that showcases the full potential of its software, but Google’s real interest lies in controlling the invisible layer of data generated by these machines.
The software operating system for robots will be just another value control point in the Google ecosystem that spans from robot workers to robot cars to robot drones and robot cameras installed on Skybox Imaging’s satellites, which record data as they orbit 185 miles above the Earth. If any
company wants to access the data captured by these machines, or program that data to get useful work done, it will need to go through a proprietary layer of software controlled by Google. And that’s where Google can extract a toll.
Even if Google decides to open-source its robot operating systems or provide open access to its robot platforms via APIs, the company will still be in position to reap the data harvest coming from each machine. Like Nest and Android before it, Google’s strategic investment in robotics is about software layers: operating systems, a set of APIs, and a suite of apps on top. How might Google use this new data asset?
> Licensing technology to other manufacturers.
> Licensing real-world maps to developers of augmented reality and virtual reality apps; in other words, providing real-world data as an on-demand service.
> Making the huge data sets available to developers through an API and reaping the windfall as more users begin to use the apps.
> Surveilling and tracking items in the real world, including tracking and recovering stolen goods.
> Facilitating virtual presence through teleconferencing and even armchair tourism via virtual reality headsets.
> Creating education and training programs set in virtualized versions of real-world environments.
These might be good for crews that must become familiar with an environment before they arrive on site.
> Allowing architects and their clients to do virtual drive throughs of a precise virtual reality rendition of the build site and its surroundings.
Dominance in AI will be propelled by three things that Google has in spades: algorithms, data, and processing power. Now it’s gathered the best minds in the field via an audacious series of
acquisitions. Given the winner-take-all dynamic in the Vaporized Economy, Google seems poised to dominate.
If you want to access Google’s intelligence-as-a-service, you’ll feed it with information of your own, thereby reinforcing Google’s winner-take-all position. If you want to use Google’s operating systems for robots (or autonomous vehicles or smart devices), you’ll need to make a similar
compromise, allowing Google to harvest the data that you generate from usage. Both of these strategies will reinforce Google’s hegemony in this emerging field.
By scooping up talent in the robotics and AI fields, Google sent a clear signal. The search giant is betting big on a future in which a vast range of human capabilities, not just physical labor but also intelligence, will be vaporized, converted into a software platform, and offered as a service. As the leading information company on the planet, Google is enviably positioned to manage the growth of this new field, and thereby transform human labor into pure information running on its software platforms.
Google’s stated mission is to organize the world’s information. The more information made available, the more profitable Google will become. It really is that simple. Every human worker possesses unique expertise in his or her particular profession, and yet that knowledge doesn’t get shared very efficiently, if at all. From Google’s perspective, that’s a tragic waste of information.
Now imagine if that information could be liberated from the human skull and turned into software:
what might happen when the best practices of half the workers in America are turned into software subroutines that can be programmed into robots? Human expertise will be vaporized: our best practices will be recorded, captured, analyzed, and turned into invisible software that can be downloaded into our devices.