LSA15: Digital Assistants and the Future of Search
April 22, 2015 | Contributed by: Joe Morsello
Digital assistants offer broader utility than conventional search. Microsoft has invested heavily and is betting on Cortana as a differentiator across screens and platforms. Mike Calcagno, Director of Engineering, Bing Experiences, gave a fascinating look at the thinking and engineering behind Cortana as well as the future of digital assistants, artificial intelligence and search.
The session kicked off with the following video:
The idea behind Cortana was to develop a personal digital assistant that knows people and helps them get things done. Utilizing analogs from real life assistants, Mike and his team have been building the tool to help users manage schedules, find information, look up directions, etc.
Typically, people aren’t inclined to utilize a natural language processing technology organically. It requires education and branding to help the tech become understood. For that reason, the team has taken the technology and used Cortana as a metaphor to embody the tool.
Cortana was a code name for the project which was taken from the Halo video game character but the name leaked. However, once it leaked, the name was “sticky” and there was some great enthusiasm so it remained.
Mike said that the personality, voice and name of the tool is important because it:
- Helps users know what to expect and understand how to approach the technology.
- Since getting people to talk to devices is not something that is habit, personality, engagement and even entertainment value helps with this.
- Over time Mike wants it to be something you recognize and trust.
Mike said that in the future, Cortana would be something that is trusted enough to accomplish things on behalf of a user from soup-to-nuts. For instance if a user needs to get a cab or book a reservation, Cortana wants to be a tool that can complete those tasks without any user management, just simply a voice command.
Another potential future state of the tool, would eliminate the need for a screen completely. Through ambient listening, users could potentially utilize simple voice commands wherever the voice recognition hardware is present.