Google Duplex and robocalls
The latest controversy
You’ve probably seen the video presentation about Google Duplex, the new assistant created by Google that can make phone calls that sound totally human interactions to the human on the other side (and to anyone listening).
If you haven’t, please go watch the 4 minutes version.
You can read more about it in this article. Not sure how this will work in real life.
But the demo looks amazing.
Honestly, I arrived to the presentation and discovered Google Duplex from the controversy in Twitter. And then I watched the video.
And at first, I didn’t even understand the problem people was having with it. (I’m not even sure I have fully understand it).
People demanded that Google Duplex should identify itself as a robot when speaking with a human. But why? I couldn’t understand.
One of the issues was that Google had also announced that it’s going to call to get the opening hours of business in a massive way. Honestly, I didn’t perceive that to be a big issue. Monthly/weekly call to get opening hours? Is that so much an issue?
So I tried to think more in terms of pros and cons. And I ended up perceiving multiple issues that are being mixed here.
But first things first
I think it is worth mentioning.
We’re taking Google’s presentation at first value. We still don’t know how good it works in real life, just a few demos.
I think it is worth pointing out the good things that Google Duplex brings.
- Good for the user. Ever since Alexa came out, there was the perception that it was useful, but not quite. Google Duplex promises to change the field.
Disclaimer: he sees danger in Google Duplex, though* However, from a local business perspective, this may be a win too. We tend to think every local service has a web page with opening hours and every restaurant or hairdresser has a booking system. But that’s not the case and in some cases it never will be. Google is effectively providing a solution that works for those businesses.
- From a technology perspective, a phone call that is indistinguishable from a human is amazing. And from my perspective an instance of a Turing test passed.
On the cons, part, I perceive the following:
- Denial of service. “If booking is easy, business may end up having loads of appointments that are later cancellation or no-show”. That’s a risk. But it is a risk that is not new. Hotels demand a credit card for making a reservation on the phone. Some restaurants don’t allow bookings in busy days at all or if not in person or if they don’t know the client.
- Monopoly. “Google is a monopoly and it’s going to be a bigger monopoly”. I won’t be the first to say that if monopolies are based on incredible service and not bullies and cronies, they may have value for society.
- Fraud and abuse. “Google has the technology to true natural language processing today. Some time in the future, many others may have it. That opens the path to fraud in massive scale”. Again, a risk that is not new. I don’t take calls I’m not expecting from phones I don’t know. Not when I was living in Spain (they used to be offering the latest broadband deal) nor since I’ve been leaving in the UK (tired of recruiters or to explain I haven’t been involved in an accident). There is nothing new in the call coming from a human or an android (not the OS :D )
- Visceral reaction. “The scary us vs them. The evil super-intelligence is here”. I’ll leave a discussion on this (and the self-identification part) for a future article. I think this is the relevant part but I perceive it is not always clear.
Yes, yes. I know not everything can be solved with technology. But I’m an engineer and I cannot avoid trying.
I remember the first thing I thought when I saw the first steps of the controversy (denial of service).
Let’s build a solution. I will create an IVR using Twilio that would say aloud opening times, and offer some options.Until I understand that those systems already exists. And if small business don’t have it, they wouldn’t buy my solution either.
But my idea “in a few years there will be human-like conversations with only androids involved”, is not original at all. In the LOL camp:
Google Duplex: "Hi! Uhm... I'd like to make a dinner reservation for 3."— Javi (@Javi) May 10, 2018
Restaurant: "Sure! What time would you like?"
Google Duplex: "it's, uhh... for tomorrow May the 11th at NULL POINTER EXCEPTION."
Restaurant: "Internal Exception: Invalid parameter not satisfying: time".
Machines talking to machines is not the only technical solution.
For the denial of service with cancellations and no-show, here is my pitch:
Cryptocurrency based appointment system. Token exchangeable for mobile credit and returned if client shows as promised.I’m launching an ICO. Do you want some coins? :D
Google Duplex seems amazing from a tech point of view.
Yes, there are risks, but in my opinion, they’re not new.
There is also the perception that Artificial General Intelligence may be possible much sooner than we expected.
PS: I decided to write this article when I got to this
Marc Andreessen is brilliant. He may have controversial opinions. I don’t always agree with some of those opinions. Benedict Evans is a similar case. But I always try to understand because they always provide something I hadn’t thought about.
Marc Andreessen, a leading Silicon Valley investor, favs a tweet comparing privacy concerns over Google AI to Jim Crow laws. 👀 https://t.co/wNtG0zE2Uz— Will Oremus (@WillOremus) May 11, 2018
Marc stopped tweeting after one of those controversies and the rage and fury (as opposed to discussion) that followed.
Now it seems he cannot even like tweets.
We should be able to exchange opinions.
First published here