If DARPA gets its way, the government won't even need to spy on you in order to record everything you say. You'll do it for them.
DARPA — the Defense Advanced Research Projects Agency — is spending $300,000 to fund the research and development of machines that can easily record and transcribe our conversations virtually anywhere.
The goal is to eventually make it possible for servers to store recorded conversations and their transcriptions from business meetings and personal interactions in a way that also makes them easy to search for.
"In their call, what [DARPA] really talked about were different areas of science where they would like to see advancements in certain problems that they see," said computer scientist Matt Lease in an interview with Wired. "So I responded talking about what I saw as this very big both need and opportunity to really make conversational speech more accessible, more part of our permanent record instead of being so ephemeral, and really trying to imagine what this world would look like if we really could capture all these conversations and make use of them effectively going forward."
"Like other AI, it can only go so far, which is based on what the state-of-the-art methodology can do," he added. "So what was exciting to me is thinking about going back to some of that work and now taking advantage of crowdsourcing and applying that into the mix."
The project, headed by Lease, will explore how crowdsourcing could help improve speech-to-text abilities in computers, smartphones and other devices. Current machines that try to do this are severely limited by their ability to understand how humans speak, construct sentences and change thoughts in mid-sentence. Even background noise presents significant problems.
Of course, regardless of technological limitations, there are obvious privacy issues concerning the project. If anyone could record anything at any time, could it actually limit the ways we interact with each other? It's unlikely anyone wants to say something that would be otherwise personal if there's a chance someone was recording it, or if there's a chance their words could land them in hot water.
Some options presented as work-arounds for the dilemma include requiring unanimous consent among participants, as well as the ability to store recordings on personal servers.
Before we can deal with personal privacy concerns, though, we'll have to welcome our voice-recognizing overlords into the world. Considering we're talking about a dream-big DARPA project here, it might take awhile.
(Edited by Lois Heyman)
