Superficially, at least, Apple’s deal for Israeli start-up RealFace (variously reported as worth $2 million to “several million”) appears not to involve marketing. RealFace has AI-based facial recognition technology that can be used to verify who you are, so is considered a cybersecurity company that does user authentication: eliminating passwords while improving security, that sort of thing.
But, looking deeper, there’s a train of thought suggesting RealFace could play a key role in Apple’s solution to a critical authentication problem related to the future of conversational user interfaces.
Think about it: Anybody in your house can order up merchandise on your Amazon Echo, right? It doesn’t differentiate who’s speaking. As chatbots become the universal UI everyone expects them to be, there will be myriad applications where you’d want to know that the user is truly authorized to do the thing he or she is doing.
More and more, AI will not only power the brains behind the conversation between marketers and audiences, but through video-based facial recognition and audio-based voice recognition it will make sure you know exactly who’s doing the talking.
BTW, this isn’t Apple’s only facial recognition AI acquisition: last year it bought Emotient, whose technology helped advertisers understand the emotional connection between advertisements and viewers by evaluating facial expressions in real time to determine attention, engagement and sentiment. Imagine a world where you don’t have to guess how the girl feels about you at the end of the first date.
Speaking of conversation-as-a-platform, Baidu’s February AI deal was for Raven Tech, which makes Chinese Siri-like AI voice assistant technology that never really took off. Raven Tech’s CEO is being put in charge of Baidu’s smart home business, and will report to Qi Lu, the Baidu COO hired in January who previously was among the architects of Microsoft’s conversation-as-a-platform vision.
Condé Nast bought CitizenNet, whose machine learning software leveraged the Facebook Advertising API to “reverse-engineer” social interactions in order to teach itself to predict advertising CTRs. That’s a bit of an oversimplification, so you may want to check out the CitizenNet CEO’s more scientific explanation from his fascinating 2011 post. Condé Nast plans to integrate CitizenNet into the Condé Nast Spire data analytics unit to expand audience targeting capabilities from the company’s own audiences to social platforms.
Meltwater acquired Wrapidity. According to TechCrunch, “Wrapidity’s AI tech can automatically figure out how to navigate web content, what that content is about, and then how to extract the content data in a structured way so that it can be interrogated for different purposes, including media monitoring. That means, for example, Meltwater will be able to quickly on-board new content sources or domains without having to manually decipher its structure and write new scraping rules.”
Ford’s deal for Argo AI does not involve media and marketing, but it’s notable in that it’s the only one with an announced price tag: $1 billion. That’s the amount Ford says it will invest in Argo over the next five years in return for its majority stake. The deal is unusual on other fronts: Argo will remain independently operated, with CEO Brian Salesky (of Google driverless car fame) in charge, will be headquartered in Pittsburgh (all the key “wetware” brains behind Argo’s AI brains hale from Carnegie Mellon University’s world-class robotics labs, including Salesky), and will incorporate Ford’s own team currently in charge of building a “virtual driver system” for Ford.
So it’s kind of a reverse-integration: instead of absorbing the company into Ford and allowing it to disappear or dissipate, Ford is throwing its own relevant talent and intellectual capital over the wall into the start-up! Ford’s stated goal is to achieve full SAE level 4 self-driving capability -- that means fully autonomous, with no human back-up -- by 2021.
HPE’s deal was for Niara, one of those cyber intrusion-detection systems. It works by establishing a “baseline” that defines a given organization’s legit cyber activity, and then looks for and investigates any activity that is inconsistent with that baseline.
For those who don’t know, that’s pretty much SOP cybersecurity stuff. The difference is that it’s all automated through the magic of machine learning, with time frames required to accomplish the work compressed by many orders of magnitude. Of course, the bad guys will just get their own software to take into battle -- this is, after all, an arms race (go reread “Neuromancer,” the novel that coined the term “cyberspace”).
And that’s all the AI M&A activity I found for February 2017; you can find the January rundown here. If you’re aware of any deals I missed, let me know in the comments below.