Microsoft Snaps Up Semantic Machines to Push Conversational A.I.
[Updated 5/21/18, 3:16 pm. See below.] Microsoft said it has acquired “conversational A.I.” startup Semantic Machines, as it tries to bolster its artificial intelligence capabilities in its longstanding competition with Google, Apple, Amazon, Baidu, and other tech leaders.
The purchase price wasn’t disclosed in a Microsoft blog post announcing the deal on Sunday. Semantic Machines had raised at least $12.4 million from investors, according to SEC filings. Its backers include General Catalyst Partners, Bain Capital Ventures, Sound Ventures, Ray Stata, and Justin Kan, per Semantic’s website. In an e-mail to Xconomy, General Catalyst’s Larry Bohn said the acquisition was a “very strong outcome for investors,” but he said he can’t share the price. [Added Bohn comment.—Eds]
Seattle-area-based Microsoft (NASDAQ: MSFT) said it will bring on Semantic’s team and establish an A.I. center in Berkeley, CA, where the startup has an office. Semantic also has an office in the Boston area. The company has more than 30 employees, according to its website.
Semantic was founded in 2014 by speech technology experts who previously worked at Apple, Nuance Communications, and Voice Signal Technologies, which Nuance acquired in 2007 for more than $290 million. Semantic CEO Dan Roth was co-founder and CEO of Voice Signal, while Semantic chief technology officer Larry Gillick was previously Apple’s chief speech scientist for the Siri voice assistant. As Xconomy was first to report in January 2015, Semantic was developing natural language processing and A.I. technology to improve “mobile voice-enabled agents.”
These days, virtual assistants and chatbots are everywhere—Siri, the Google Assistant, Amazon’s Alexa, Samsung’s Bixby, Microsoft’s Cortana, and more. But their capabilities are still primitive, mostly limited to simple commands and questions, like setting a timer for baking cookies or asking about the weather forecast.
Virtual assistants “aren’t able to understand meaning or carry on conversations,” Microsoft said in the blog post. “For rich and effective communication, intelligent assistants need to be able to have a natural dialogue instead of just responding to commands. We call this ‘conversational A.I.’”
Microsoft said Semantic is making progress on this front with machine learning technologies that can “enable users to discover, access, and interact with information and services in a much more natural way, and with significantly less effort.” Microsoft envisions Semantic building upon Microsoft’s own A.I. advances, which include developing a speech recognition system that in 2016 demonstrated the ability to match human transcriptionists for accuracy.
The stakes are high; some experts see voice recognition as one of the most important next frontiers in human-machine interaction. That has implications not only for the way people access information, buy products, and interact with each other online, but also for the future of work and society.