Google 2018 I/O Keynote: Google Assistant will make phone calls for you

[RiverSong]

Well, this is interesting. I watched a recap of the Google 2018 I/O keynote and I have to admit, I was shocked. A lot of big announcements were made by the tech giant. Things like a new a "Android update on the way. John Legend is lending his voice to Google Assistant. Gmail can almost write emails entirely by itself. AI was a big theme throughout." (The Verge)

Here's the video recap:

The one that stood out the most to me was Google's new Duplex technology. According to The Verge:

"Onstage at I/O 2018, Google showed off a jaw-dropping new capability of Google Assistant: in the not too distant future, it’s going to make phone calls on your behalf. CEO Sundar Pichai played back a phone call recording that he said was placed by the Assistant to a hair salon. The voice sounded incredibly natural; the person on the other end had no idea they were talking to a digital AI helper. Google Assistant even dropped in a super casual “mmhmmm” early in the conversation."

You can read the rest of the article:
Google just gave a stunning demo of Assistant making an actual phone call

There's also Gizmodo's article with an interesting title:
Uhh, Google Assistant Impersonating a Human on the Phone Is Scary as Hell to Me

This tech is still under development with "early testing of Duplex inside Assistant this summer." So, I'm curious. What are your thoughts about these announcements and most importantly the Google Duplex? Is it something you'd be interested in using?

 

[Fridgemagnetman]

The amazing thing is that Assistant can actually understand the nuances of conversation,” he said. “We’ve been working on this technology for many years. It’s called Google Duplex.”

For such tasks, the system makes the conversational experience as natural as possible, allowing people to speak normally, like they would to another person, without having to adapt to a machine.”

Google AI seems to be working on 'small talk'

So the sterotype is the autistic person gets a job with computers.

To improve it's programming Google AI will have to talk with people good with 'small talk'
Therefore,in future :

Computers will also ignore us.

 

[RiverSong]

Well, I certainly hope that's not the case. To be shunned by AI would be something, lol.

Personally, I don't use Google Assistant. But if this tech becomes of real practical use then I might consider it.

I mean, having an AI participate in small talk in place of a human sounds great. It would be a good time/stress saver. As a person who doesn't like and often finds it difficult to have phone conversations, I would be happy to have an AI assistant. Of course, there are some privacy vs convenience tradeoffs, I'm sure.

What I'm wondering is if this becomes widely used and accepted in the near future then would the days of human-to-human phone conversations become obsolete. It would seem like humans would get cut out of the process entirely and just AI would be speaking to other AI.

This could all get pretty sci-fi...

 

[Fridgemagnetman]

Well, I certainly hope that's not the case. To be shunned by AI would be something, lol.

Personally, I don't use Google Assistant. But if this tech becomes of real practical use then I might consider it.

I mean, having an AI participate in small talk in place of a human sounds great. It would be a good time/stress saver. As a person who doesn't like and often finds it difficult to have phone conversations, I would be happy to have an AI assistant. Of course, there are some privacy vs convenience tradeoffs, I'm sure.

What I'm wondering is if this becomes widely used and accepted in the near future then would the days of human-to-human phone conversations become obsolete. It would seem like humans would get cut out of the process entirely and just AI would be speaking to other AI.

This could all get pretty sci-fi...

I've started using it breifly. To open apps and stuff.

I think I would use it for appointments and stuff, but a lot of them you can make online. The voice typing is also quite good, but the erros can be bad enough that you didn't know what you said when you go to correct.

I mean, having an AI participate in small talk in place of a human sounds great. It would be a good time/stress saver.

But then would the AI gossip about us

The robocalls in America are also almost impossible to tell if they're a human or not.
Often you can find yourself answering "yes" which,in case of a phishing scam, is all they need, is your voice saying yes.

One reason to keep living is just to see what happens with all this stuff.
Sci first becoming real every day

 

[pjcnet]

Voice recognition has still got a long way to go and compared to other computer technology hasn't improved anywhere near as much since the 1980s (I've used voice recognition even back then), if the AI is going to have an interactive phone conversation it's going to have to be a lot better than many of the automated voice recognition bots we currently have on the phone where we need to speak super clearly and carefully for it to pick it up properly, and they only have to listen to a choice of answers which is much easier than having to potentially recognise anything that is commonly used in the English dictionary as well as names Etc. with a virtually infinite combination of possible sentences in multiple dialects and expressions, but it will need to recognise voice emotion too for it to work really well and react appropriately to it, something I've never seen done so far with voice recognition. The same sentence or expression can mean completely different things depending on the way you use expression, for instance I could say "get lost" as a friendly joke or "get lost" in a very nasty way, could a computer tell the difference?

The hair cut appointment is very well and good in theory and for demonstration purposes where everything goes smoothly and perfectly with the woman answering clearly and predictably exactly as the AI expects her to, but this often isn't what happens in real life. What happens if the woman on the other end has a strong accent and/or doesn't answer exactly as the AI expects? Perhaps she could say, "please hold" while she's dealing with someone else or what if she started asking about whether there was a particular person you wanted to see? Perhaps she might ask specific questions about what's required to enable her to choose the best person to perform the job and there could be a choice of appointments, she could ask whether you prefer a morning or afternoon and all that has to be understood and interpreted properly by the AI or it will have to respond with something stupid like "I'm sorry I don't understand, please reword your question". What if she started chatting about for instance how busy they've been lately or just general friendly chit chat? What if she asked any question that the AI isn't programmed to respond to, it could even be something like would you be happy to do a satisfaction survey for us, but there's virtually an infinite number of unpredictable possibilities? People are very unpredictable in real life and the AI has to take into account every possible situation or it will fail miserably and give itself away. Even if the AI bluffed it's way through this, what would happen if the AI called and another automated system answered with a voice recognition bot? It would almost certainly fail unless it was a compatible system that was specifically accounted for. There's masses of things that can go wrong and often it will go wrong with such a system. It could however be much worse if the person on the other end doesn't realise that an AI bot has called them in your name when it goes horribly wrong, it could cause awkward and embarrassing situations for yourself, would you really trust such a system? At this point in time it's a "now way" for me.

It might work with some people, but usually the person at the other end will know they're talking to a computer after they try to respond to the initial conversation with anything but a perfectly predicable clear response and they may not be comfortable talking to a machine. It will be common for the computer to misinterpret the voice and respond inappropriately with current technology, this will be infuriating for many people. How often have you called a company and you only want to speak to a real person, but instead you have a dumb computer voice recognition system that won't put you though and won't interpret what you really want? Well when a computer calls you there's unlikely to be an option to talk to a real person and a lot of people will end up putting the phone down in fury unless it's just a simple message with little interaction required. It will get better, but based on how gradually voice recognition has improved so far I suspect it will about 25 years before it will be almost impossible to truly tell that you're talking to a computer in an unpredictable two way conversation. AI itself can be made pretty good even now if it recognises and interprets the voice properly and it takes expression into full account too, this has been partly proven with typed text only two way conversations with an AI where people haven't been able to tell, although there's more potential for even this to go wrong when the computer is trying to achieve something very specific from the conversation, also with text only there's no voice expression. Even when the voice recognition and AI works perfectly in every way, many people will still be uncomfortable talking to a machine, although as years go by this could also become more accepted in younger generations.