When thinking about the Pentagon’s technological research, it’s more pertinent to wonder what its scientists aren’t into than what are they into. The latest doozy from its James Bond-like Q-wing, the Defense Advanced Research Projects Agency, is a real imagination stoker: total transcription of every conversation on earth.
Yup. According to Wired, DARPA has given University of Texas computer scientist Matt Lease $300,000 over two years to work on a project called, “Blending Crowdsourcing with Automation for Fast, Cheap, and Accurate Analysis of Spontaneous Speech.” As per the article:
The idea is that business meetings or even conversations with your friends and family could be stored in archives and easily searched. The stored recordings could be held in servers, owned either by individuals or their employers.
This would be accomplished with all sorts of recording devices, from smartphones and video cameras to audio recorders. Algorithms would process the conversations, then upload and store them for future use.
One of the issues, however, would be marshaling humans to essentially copy edit the results. Machines could conceivably do the heavy lifting, but they still tend to miss things like punctuation, nuance and spellings of names. Since it is the military looking into this, having human shepherds could work in intelligence settings, where the volume of conversations being transcribed would be relatively small.
But in order for this to be applied to the larger civilian world, humans would have to be taken out of the equation entirely. Any journalist knows that transcribing conversations is the dullest job around, so it’s hard to imagine anybody willing to do so on a large scale. Even if someone were willing to pay them tons of money.
If it could be entirely automated, the implications would be staggering. For one, we’d have total recall of every conversation we’d ever had. No more fighting with the wife or girlfriend over what you or she may or may not have said. Let’s go back to the tape.
Search engines could mine our everyday conversations, with a bevy of potential uses springing from that. How about speech-based ads—“hey, we heard you talking about margarine just now—have you seen our new butter?”
To those same transcribing journalists, this technology seems like a godsend. No more hours spent going over recordings of interviews—it’s all instantly done for you.
But wait a minute, not so fast. Combine this sort of technology with robots that can write and we’re one step closer to being obsolete. If robots can figure out how to ask good questions, there really won’t be much need for human journalists. Then we can really start lamenting the state of the business.