Apple has acquired a start-up that uses artificial intelligence software to analyse emotions from facial expressions, the latest in a string of acquisitions that may point to a push into virtual or "augmented" reality or enhance its automotive research.
San Diego-based Emotient was founded in 2012 and has already been awarded patents covering how it uses crowdsourcing to "train" its machine-learning technology for sentiment analysis and — crucially for the privacy-conscious Apple — a way to understand facial expressions without storing recognisable images of individuals' faces.
Blog posts on Emotient's website, which have been removed following Apple's acquisition, described how it "harvests faces" to create "emotion-aware machines". A team of people, via crowdsourcing, was used to train its machine-learning system by collecting and labelling the emotions in 100,000 images a day, "quickly converting human intelligence into artificial intelligence", Emotient said in a press release last May.
AI has become a huge new focus for Silicon Valley companies lately. Earlier this week, Facebook chief Mark Zuckerberg said he planned to build an AI-based virtual assistant akin to Jarvis in the Iron Man movies and comics.
However, Apple has been seen as lagging behind Alphabet, Google's parent, and Facebook in its AI capabilities. Apple's hardline stance on protecting its customers' privacy by not storing or using large volumes of data for features such as personalisation has also been seen as a potential impediment to developing machine learning.
A promotional video released by Emotient two years ago asked: "What if your devices could read your emotions and respond to them?" One example showed a car navigation system suggesting a new route based on detecting that the driver was frustrated.
With fewer than 50 employees according to its LinkedIn profile, Emotient had raised about $8m from investors including Intel Capital. Its early customers included marketers and retailers that wanted to understand how consumers were responding to their products. Last year, it released a Google Glass app that could be worn by salespeople, for training and to analyse customer reaction in real time.
Emotient had also suggested its technology might be useful in education or even medical diagnosis. However, Apple often makes technology-focused acquisitions that end up being used in entirely different applications to when the company was independent.
In August, Emotient trained its technology on a Republican presidential candidate debate and concluded that while Donald Trump "predominantly conveyed anger", Ted Cruz "almost exclusively expressed sadness" during the Fox News broadcast.
Apple confirmed the acquisition, which was originally reported in the Wall Street Journal, saying it "buys smaller companies from time to time and we generally do not discuss our purpose or plans".
While Emotient's technology might potentially be used in Apple's own retail stores, which are undergoing an overhaul under former Burberry chief Angela Ahrendts, the acquisition follows deals in the past year that some analysts believe point to some kind of VR or "augmented reality" system, where virtual images are intermingled with the real world.
Last year, Apple patented a VR headset akin to Google Cardboard or Samsung's Gear VR that uses a smartphone as a screen. In May, Apple acquired Metaio, which used a smartphone or tablet to overlay digital information over a real-world image. In November, it emerged that it had acquired Swiss-based Faceshift, a maker of motion capture technology that maps facial movements to digital characters in movies such as Star Wars: The Force Awakens.
Analysts at Piper Jaffray concluded that Apple "has a team exploring the AR space". "We believe Apple's early involvement in the space suggests the company is preparing for the next evolution of computing," it said in a note last year.
AR and VR were big themes of the Consumer Electronics Show this week in Las Vegas, as were transportation and autonomous driving.
Apple has a team working on automotive technology with an eye to releasing a vehicle in the coming years, the FT and others have previously reported. After poaching employees from the likes of Tesla, its executives met California regulators in September to discuss rules around testing driverless cars. Technology such as Emotient's, Faceshift's and Metaio's could be used to help build a car with new kinds of heads-up displays or provide the ability to respond to messages and notifications hands-free while driving.