‘You don’t sound well today…’
The Washington Post’s A.I. correspondent (he’s a human covering A.I., not the other way around) yesterday tweeted out a patent application
from Amazon. The technique they hope to patent is “an Alexa system that can tell when you’re sick (so it can sell you cough drops). Technical description: a voice-processing algorithm that detects when a user’s emotional state is abnormal.”
I thought about this this morning as I snuffled with a cold, asking my Echo to stream my favourite radio station. If Alexa had responded “you sound bunged up – should I order some cough medicine on Prime Now?,” I’d have probably said yes. It might have been delivered to my door by the time I wrote this sentence a couple of hours later.
The obvious response to technology like this is to recoil in horror. But it’s hard not to feel a sense of delight when tech gets something totally right without you asking. Google Photos recently sent me – unprompted – a montage video called 'They Grow Up So Fast,’ showing my daughter as a baby, growing into a toddler over the space of several images. Yes, I was a little creeped out, but it was lovely to watch, too.
The potential for this tech to be used for the wrong purposes is undoubtedly a serious concern. What if Amazon tracked how vulnerable you sounded, and then offered you unbeatable deals on things you didn’t really need when it thought you were most likely to say yes? And what if Google’s tech was sold to law enforcement so they could collect every photo of each of us all ever taken, just in case they needed them as evidence in the future?
That’s why A.I. ethics boards at tech companies are so important, but given how prevalent this kind of tech is becoming, we’re going to have to learn to take the good with the bad. I just hope the bad uses don’t impede the benefits pre-emptive A.I. can offer.