Google and Facebook May Own your AI-Enabled Future

Google, Facebook and Amazon are way ahead of the curve when it comes to creating AI products and using the technology to power their wares. Josh Ong, Director of Global Brand Strategy & Communications at Cheetah Mobile, explores what this means for AI startups, who will be compelled to increasingly rely on these giants to provide the AI resources and capabilities to power their endeavors

Artificial intelligence is one of those things that simultaneously seems to be a long way off from making a noticeable impact in people’s lives while also well on its way to actually doing so. In other words, it hasn’t quite lived up to its potential yet – but there are companies who have come up with innovative and ingenious solutions using AI. One has to look no further than one’s mobile phone to understand the impact that machine learning and deep learning have already had on our day-to-day routines, through personal assistants such as Siri and the algorithms that power a phone’s ability to do everything from recognizing the faces of friends and family to customizing apps based on a user’s personal preferences.

And yet, the names behind these AI achievements are eminently recognizable, with Google, Apple, Facebook, and Amazon leading the charge. By and large, there have been relatively few AI startups, who have been able to make a lasting impact on the AI world – and those that do, often find themselves quickly snapped up by these leading companies.

Big companies have also largely cornered the market on AI talent, recruiting heavily from universities with AI departments and poaching well-known luminaries from various institutions. This naturally has created a problem for other companies looking to hire professionals with AI experience, as not only are there relatively few people with those qualifications to begin with, but most companies cannot even hope to offer the kinds of salaries that those professionals can command. The New York Times reports that even newly-graduated Ph.D.s and people with only a few years of experience can command salaries ranging from $300,000 to $500,000 including company stock – an astronomical amount for people with limited job experience. Some have compared the experience to the process for signing top athletes, joking that the tech industry “needs a National Football League-style salary cap on AI specialists.”

This puts a considerable wrench in the plans of any company looking to enter the AI space, big or small. After all, few companies can match the $120 million in incentives Google paid to former employee Anthony Lewandowski before he left the company to found his own startup. And, it’s not merely a matter of finding the right talent. Building an AI platform requires other resources – at the very least, a considerable amount of computing power, and access to the data required to train the AI.

Here again is where large companies such as Google and Amazon show their dominance. Not only do they develop AI products for their own use, they are also willing to give others access to (some) of the capabilities they have already developed – for a small fee, of course. Google’s cloud platform offers “fast, large-scale and easy to use AI services”, including a large-scale machine learning service, the ability to build your own chatbot, and image and speech recognition. Amazon and Microsoft also offer their own machine learning services for developers, giving them yet another chance to assert their primacy in the AI marketplace.

And, it’s not just developers who are increasingly relying on companies like Google, Facebook and Amazon to provide them with AI capabilities. Most of the AI-enabled products we use every day, from Alexa to Facebook tagging to the algorithms used to flag extremist content, have been developed by these same companies. Moreover, the growing popularity of voice assistants means that any company looking to reach customers via Alexa, for example, will have to rely on Amazon’s platform to provide the voice capabilities and the access, thus limiting the need for companies to develop their own voice recognition or chatbot abilities.

Despite all of the considerable barriers to entry that exist for any company looking to enter the AI space, the number of AI startups and AI products on offer has continued to climb. AngelList lists almost 4,000 AI startups with an average valuation of $4.9 million, which indicates the level of competitiveness in the industry, as well as the fact that, despite their best efforts, Facebook, Google and Amazon have not yet managed to completely take over the market for AI.

What the big companies have going for them, however, is the amount of data they have access to, and the amount of resources that they can continue to throw at AI development. There’s no getting around the fact that there are significant barriers to entry for smaller companies looking to find opportunity in the AI market, and that most companies will find themselves relying on the giants in some way, whether to provide computing power or the ability to build a machine-learning model. This doesn’t mean that they won’t have the opportunity to create a fantastic AI-enabled product; it just means they won’t be able to go it solely on their own.