Teaching machines to understand human language has long been the holy grail in artificial intelligence – and in a Montreal startup, Microsoft Corp. believes they have found the perfect guide.
On Friday the Redmond, Wash. tech giant announced that was acquiring Montreal-based deep learning startup Maluuba, which Harry Shum, executive vice president of Microsoft’s Artificial Intelligence and Research Group, called “one of the world’s most impressive deep learning research labs for natural language understanding,” for an undisclosed sum.
“We’ve recently set new milestones for speech and image recognition using deep learning techniques, and with this acquisition we are, as Wayne Gretzky would say, skating to where the puck will be next — machine reading and writing,” Shum wrote in a Jan. 13 blog post. “Maluuba’s expertise in deep learning and reinforcement learning for question-answering and decision-making systems will help us advance our strategy to democratize AI and to make it accessible and valuable to everyone — consumers, businesses and developers.”
In a Jan. 13 blog post of their own, Maluuba cofounders Sam Pasupalak and Kaheer Suleman called the acquisition an “important milestone” for their company, and said their team was “incredibly excited” to be working with Microsoft.
“Their ambitious vision of democratizing AI to empower every person and every organization on the planet fundamentally aligns with how we see our technology being used,” Pasupalak and Suleman wrote.
Though its most notable product to date has been an Android-based alternative to Siri, Maluuba’s goal since being founded in 2012 has been to create a deep learning platform modelled after the human brain. Rather than the text- or graphic-based pattern-matching of many current AI platforms, the company’s team has dreamed of building systems that can comprehend, synthesize, infer, and make logical decisions like humans, Pasupalak and Suleman wrote.
Thus far, they wrote, Maluuba has focused on machine reading comprehension, understanding dialogue, and human intelligence capabilities such as memory, common-sense reasoning, and information seeking. The company’s early achievements in these areas has required the founders to scale their team rapidly, making Microsoft’s acquisition offer especially welcome.
By joining Microsoft, Pasupalak and Suleman wrote, Maluuba’s team not only has access to Microsoft’s immense technical resources, but the opportunity to deliver its work to the billions of enterprise and consumer users who stand to benefit from the arrival of truly intelligent machines, a sentiment echoed by Shum.
“Imagine a future where, instead of frantically searching through your organization’s directory, documents or emails to find the top tax-law experts in your company, for example, you could communicate with an AI agent that would leverage Maluuba’s machine comprehension capabilities to immediately respond to your request,” Shum wrote. “The agent would be able to answer your question in a company security-compliant manner by having a deeper understanding of the contents of your organization’s documents and emails, instead of simply retrieving a document by keyword matching, which happens today.
“Maluuba’s impressive team is addressing some of the fundamental problems in language understanding by modeling some of the innate capabilities of the human brain, from memory and common sense reasoning to curiosity and decision making,” Shum continued. “I’ve been in the AI research and development field for more than 20 years now, and I’m incredibly excited about the scenarios that this acquisition could make possible in conversational AI.”
Both Shum and Maluuba’s co-founders said they would have more to share regarding Maluuba’s assignment for Microsoft within the next year. In the meantime, Pasupalak and Suleman’s engineering and research team will be joining Microsoft’s Artificial Intelligence and Research organization.
“Much like its overall vision, Maluuba’s approach to research and engineering is also directly aligned with ours,” Shum wrote. “Part of that approach is contributing to the research community for the common advancement of AI systems. For example, last month we released the MS Marco dataset for machine reading; Maluuba also recently made its datasets for reading comprehension and conversational dialogue systems available.”
