Earlier this week, Google kicked off its developers conference, I/O, and showed off quite a bit. There were new features for Google Maps, Android, and plenty of other things to talk about. But, as you might have already seen by now, the most talked-about part of the keynote speech was showing off Google Assistant’s ability to make a phone call and interact with a human being.
Just talking to a human being isn’t all that eye-popping. The digital personal assistant does that every day already. But it’s the unveiling of what Google calls Duplex, and Assistant’s ability to have a *convincing* conversation with a real person that stood out. Assistant wasn’t just saying a simple sentence. It was having a conversation, with pauses and additional words to make it seem more realistic.
Basically, Google made it so that in specific situations Google Assistant can sound, speak, and react like a human being.
Let’s not beat around the bush here. Getting to a point where artificial intelligence sounds like a human being is the goal. Whether that’s a good goal to reach has always been a hot topic, especially for science-fiction writers. And while Google Assistant, Siri, Cortana, and Alexa all sound “human enough”, their artificial roots are still obvious.
The concerns have been aired by many people already. Should Google be making something like this? Some people think Duplex is too far because it aims to sound human and therefore trick the person on the other end of the line. A conversation based on a lie, basically. And some folks think it’s unfair for the person on the other end of the phone, especially those in the service industry, as they are the brunt of a conversation that’s one-sided in its human connection.
Truthfully, the concerns are valid. And we all knew that, one day, we’d be having to address them. But some folks thought that might not be so soon. But Google took the stage, and we watched as a digital personal assistant spoke to a real person and set a salon appointment.
It was pretty awesome.
The feedback was positive and negative. I personally think Duplex might be worthwhile one day, especially for people who aren’t particularly fond of talking to other people. Even over the phone. But then I also don’t like the idea of being cold called by robots. Just imagine political campaigns with the power of Duplex, and that sounds legitimately awful.
Google’s Sundar Pichai didn’t make any mention of the fact that Duplex would let the person on the other end of the line know they were talking to an AI, and that developed concerns, too. Because a lot of people think Duplex, or any AI for that matter, should tell us that’s what we are talking to.
Duplex will have disclosure built-in, according to Google. So, in some way or another, if Duplex ever rolls out broadly people will be told they are talking to a robot. Which leads me to ask a very simple question: How long are those conversations going to last?
It sounds like Google is basically making a way to test just how long some people will stay on the phone with an AI. Which leads to more questions. For one, if Duplex tells someone on the phone that it’s Google Assistant they are talking to right off the bat, is that grounds to just hang up? And if Assistant doesn’t tell the person on the other end, in very plain language, that they are talking to an AI, that’s not disclosure at all.
So if Google wants this to work — if they see a genuine future with Duplex — then they will need to get over the growing pains. They will need to show that Duplex is genuinely helpful, while also offering that disclosure right out of the gate.
Maybe it works out. Maybe this is just one more Google creation that never really gains traction and we all talk about it one day in the future as a distant memory.
Where do you stand on Duplex as it exists right now? Are you excited to see what it becomes? Or would you be happy to see it disappear?
Like this post? Share it!