Artificial-intelligence stocks can feel like blockchain stocks: Every company wants a piece of the action. Results are what set the contenders apart from the pretenders.
How do you measure results? There's no perfect method, but patents applied for and issued tend to indicate a level of seriousness and financial commitment that's hard to refute. In 2017, IFI CLAIMS Patent Services found these five companies to be the most provably inventive in the category of "machine learning," a style of artificial intelligence in which systems scan huge volumes of data for patterns that represent opportunities to be exploited.
Why are they investing so much in AI? Is what they're researching potentially game-changing for these businesses? Read on for some answers.
IBM: Making Watson more like Sherlock
Published machine-learning patents last year: 654.
For a taste of what's to come: Read patent application US20170323075A1.
IBM has been in the AI business in some form since before the Deep Blue chess-playing computer got the best of then-world champion Garry Kasparov 21 years ago. So there's good reason to believe IBM is serious about making real machine-learning-powered products from its intellectual property.
In the case of US20170323075A1, IBM is proposing using machine learning and predictive models to better understand the health risks you or I might face based on a wide variety of factors, using algorithms to calculate the odds. You can bet Watson will be involved if this patented idea is ever brought to life as a product.
Microsoft: Don't hang up yet
Published machine-learning patents last year: 139.
For a taste of what's to come: Read patent application US9785174B2.
Microsoft has good reasons to be working on machine learning and AI -- such as Bing. So what if it's the second-tier search engine? It's still a massive traffic driver, and source of intelligence for what customers and potential customers may want or need. And that's just one application.
Patent US9785174B2 speaks to using predictive models to determine when to moderate the carrier signal strength transmitting into a smartphone, so as to not overexpose humans.
And its subsidiary LinkedIn? The networking company published 70 machine-learning patents last year.
For a taste of what's to come: Read patent application WO2017147356A1.
LinkedIn may be a part of Microsoft now, but it's still a substantial business with a need to make better use of data.
In this patent, the business network uses predictive analytics to help recent college graduates narrow their job search, by showing where their attributes might play best; for instance, revealing where fellow alumni work can be helpful. The bigger leap LinkedIn proposes is examining listings to determine whether a job requires or desires prior experience -- an important distinction for recent grads in search of their first work experience.
Google: Patenting Nostradamus
Published machine-learning patents last year: 127.
For a taste of what's to come: Read patent application US20170308539A1.
The company that wants to organize and optimize the world's information is using machine learning and AI to find new ways to do it.
Just look at the patent. Google's proposal for "predictive generation of search suggestions" is straight out of sci-fi. Using your device's location and search history, Google presumes to know ahead of time what it is you're likely to be searching for. Talk about fascinating, and scary: Were this to come to every phone that uses Google -- which may as well be all of them -- you could expect the company to keep trying to read our minds.
Facebook: Please like more stuff
Published machine-learning patents last year: 66.
For a taste of what's to come: Read patent application US20170324820A1.
No surprises here; Facebook's patent activity is geared toward figuring out ways to get you to like more stuff.
In this particular patent application, filed in November, the company describes a method for interpreting what you might like to make better recommendations using "implicit interactions." So, for example, a user who visits a new page or user profile for the first time without liking it is considered to have performed an "implicit interaction" that Facebook can then use to recommend other pages or place an ad in the news feed. There's a creepiness factor here, to be sure. Why patent such an invasive feature? Facebook, and other data-driven businesses like it, may not have much choice. Every data signal is key to successful advertising, which is key to growth.
Patents aren't perfect windows into the future, and AI as a technology for business use is still in the formative stages. But as the market shifts and the opportunity becomes more clear, it's a good bet these five companies will be leading the pack.
Suzanne Frey, an executive at Alphabet, is a member of The Motley Fool's board of directors. Teresa Kersten is an employee of LinkedIn and is a member of The Motley Fool's board of directors; LinkedIn is owned by Microsoft. Tim Beyers owns shares of Alphabet (A shares) and Alphabet (C shares). The Motley Fool owns shares of and recommends Alphabet (A shares), Alphabet (C shares), and Facebook. The Motley Fool has a disclosure policy.