Google Assistant making real phone calls?! Artificial intelligence is taking crazy steps forward. How far will it go?
Talking AI with Neil Degrasse Tyson: https://youtu.be/y9U10qdDQqk
Google Duplex blog post: https://ai.googleblog.com/2018/05/duplex-ai-system-for-natural-conversation.html
MKBHD Merch: http://shop.MKBHD.com
Video Gear I use: http://kit.com/MKBHD/video-gear#recommendation17959
Tech I'm using right now: https://www.amazon.com/shop/influencer-0bfe542e
Why don't they do it so that the assistant announces that it's an assistant calling at the top of the greeting? Seems acceptable and it avoids the creepy issue. When people get used to the assistants they won't hang up on them. (Update: ok you talked about it towards the end of the video).
I don't know if I believe this is real, honestly. It's clearly in a controlled environment and I feel like it's going to flop when it's released, honestly. I think they should market this more towards your assistant to sound more like a person than something that's this advanced. Just my opinion though.
everyone would need to present themselves as a human or robot in phonecalls if we want it that way then.
its a really weird time we are living in today. dealing with human v android interactions. lawyers and engineers are going to be making a bunch of money in the upcoming years ill tell you that.
This is great. Now when we call a company, we have push through a maze of options until we reach the correct department and then put on hold. Soon Google assistant will go the maze and patiently wait, so we don't have to waste our time. The tables have turned. The consumer is in control.
There will always be skepticism and concern from simple-minded individuals when a specific field of technology reaches a new zenith. Rest assured, once these new AI features are implemented and functioning at a high rate, most of the doubters will eventually come around. Also, informing the people that they are speaking to a robot is exactly like telling someone the vehicle next to them is self-driving. It's unnecessary as long as that device is functioning as intended. Moreover, informing other humans would only cause some of them to behave abnormally and unpredictably (like staring at the vehicle, trying to cause accidents for insurance frauds, testing the AI range of responses, exploiting known faults, or hanging up) which leads to poor performance/results of the AI or autonomous vehicles.
NO! NO! NO! Sorry this will excite crims everywhere. We already have voice recognition. Just add voice recognition and record and then change the voice to the person recorded. Now anybody can use your voice to represent they are you through Gooogle. Get over the WOW factor IT people... This is fraught with danger and crims are always one step ahead of the good guys. Wake up we don't need this!
software development is evolving from coding into AI training/growing/maturing, in a few decades, "do you remember how 50 years away people had to type every single instructions and actually think about shit going worng? No man, they mess it up a lot, they called that bugs"
"if you can't tell the difference, does it matter" -- Westworld hostess (played by Elon Musk's ex)
Umm, this comment is definitely from a human. I's appreciate your effort at making this transition easy. We's shared your link on all the things. Um, not like weirdly on every single thing possible within less than a second but just like on I's twittergrams. You know?
You are doing a good job. Goodbye bye.
Why not just have google duplex on the service side. A restaurant doesn’t have to waste time answering the phone. Or maybe when you call tech support for your ISP or your mobile provider and hen at least we aren’t waiting on hold for 1.5 million hours?
I think it will be good, as you said at times we may not have the time to call ourselves because we are off to work or something more important. I really don't think we need to know we are talking to a robot because if we do then the conversation we take the wrong turn or defeat the purpose of the call. I know it might be a little concerning at first but so we assistant and al other forms of AI when they came out. As you said though, it all depends on how far the goal posts will go.
Next steps: 1) Google trains Duplex to speak with YOUR voice. 2) Google asks to build up a knowledge profile of info contained in public and private media that you've either consumed or produced, so Duplex can respond increasingly contextually as YOU. 3) Google encourages you to use life-tracker tech so that Duplex will become current and converse even more like you over time. 4) Google allows you to deploy multiple Duplex "agents" out into real and virtual worlds, doing your bidding and only checking in with summaries or when it has some threshold of uncertainty of how you would respond/behave in a given context. 5) In years to come, your Duplex agents will occasionally inhabit compatible telepresence robots to give YOU physical presences elsewhere while you're doing something completely different.
Some of the ways I see adaptations of this tech being misused:
- army of sales bots
- army of con bots.
- massive security threat combined with the upcoming voice imitation technology.
- stalker army released to gather input from the victim and people around them.
The list of course goes on and on, just like the list of positives. I'm just not certain if we can keep trusting phone based conversations, which in the end just might mean that the product will fail to bring value for society as a whole (when also considering negatives such as loss of jobs). Thank God that we can still trust video calls... for now :)
Boss you haven't looked at the wider goal post. What's going to happen next is assistant is going to learn your voice and use that to make appointments. At least now there are only a few voices that google can use and that makes it easy for people on the phone to recognize over time that this is human and this is robot. However, when assistant starts to learn the voice of a person, that distinction can not be made.
technology in the wrong hands, will make things worse. today I have more than 30 unwanted robocalls on my phone. if the bad guys have the google duplex technology, I just cannot imagine...
I think law enforcement must walk hand in hand with new technology.
great topic, great video, by the way.
I think it would be much more useful if the Google assistant would be on the hairsalon's or other small businesses side. Like giving out appointments is not hard and that would definitely save a lot of time! I could be easily working out of a google calendar as a database and at the end of the day manager could check out what is going on with appointments
Unless we're seriously going for universal basic income I'd be awful worried to be in any service industry right now. Also you know it's only a matter of time before companies are using this technology instead of those standard robo calls to solicit you.
I'm shared between the feeling that's an amazing idea, to don't waste time for this kind of phone calls, and the beginning of our end, in a kind... I think we should be careful about the next of the AI technology... and listen Elon Musk about it... We shouldn't push the "AI button" to much and being unable to backtrack... it could be dangerous... I wouldn't want that the Science Fiction Movies like iRobot or Terminator, have been prophecies.... It's amazing to invent and create science fiction movies, and sometimes to try it for real, but don't let our craziness about technology and AI, being too much "Real" ... It can be dangerous.
Pass the Turing test? *No* ! The Turing test specifies that the person *has to know* that he might be talking to a machine and still be unable to distinguish between that and a real person even if he made an effort (like asking very specific things). There is also no time limit. The person could call every day and ask the potential AI about its childhood or whatever for months before making up their mind. If this would classify as a pass, then the prerecorded messages of people cursing and yelling which we used in the 90s for prank calls would also qualify, as in half the cases the person being pranked would get angry and hang up.
I see businesses using this to replace their overseas customer service reps. I know I hate talking to them over in Bangladesh, India, The Philippines, etc... Not because of where they are but because they have trouble understanding what I am asking them or telling them 9 times out 10. This might reduce the amount of misunderstandings and get you to the right department or to the correct live person quicker. Yes more people will lose jobs, but that has already been the case. Look at supermarkets and pharmacies like CVS & WALGREENS. They have self checkout lines. Amazon has grocery stores with no cashiers. This is both a good and bad thing. Good for Businesses and Bad for Employees.
first, i wouldn’t mind not knowing i’m talking to a robot.
second, we already have SO many robots or “robots” helping us in our daily lives. especially if you use the standard definition of robot (a machine capable of carrying out a complex series of actions automatically).
Google Duplex is just going to further this and make people’s lives easier