This is a pretty common pattern. Lots of websites also have &quot;establishing secure connection&quot; interstitials for the same reason.

That famous tax preparation software added several screens to &quot;review&quot; the data.

It&#x27;s really just an audio version of a loading bar or spinner - users get really uncomfortable if the UI becomes unresponsive for even a few hundred milliseconds, but they&#x27;ll wait for several seconds if it looks like something is happening.See also:<a href="https:&#x2F;&#x2F;en.wikipedia.org&#x2F;wiki&#x2F;Comfort_noise" rel="nofollow">https:&#x2F;&#x2F;en.wikipedia.org&#x2F;wiki&#x2F;Comfort_noise</a>

I&#x27;ve heard the HP12c calculator also slows down its screen refresh on purpose because people couldn&#x27;t believe the math was right when it first came out and it was blazing fast.

Most of the flight search companies do the same (&quot;Finding the best&#x2F;cheapest flights for you&quot;). It&#x27;s almost instantaneous, but they introduce this artificial wait.

It&#x27;s not a new thing. A famous tax preparation software introduced a &quot;compute&quot; screen that took a few seconds to make people more comfortable with the results even if the computation itself is instantaneous.

The &quot;um&quot; isn&#x27;t a deficiency, but the slow response is. If the response is artificially delayed to give the appearance of slow thinking, and an &quot;um&quot; added to fill the artificially long silence, that&#x27;s an artificial deficiency.

I don&#x27;t know if I&#x27;d call it a &quot;deficiency&quot; - if we interpret &quot;disfluency&quot; in a literal sense as &quot;not flowing&quot; without negative connotation, then the interruptions (hmm, uh, okay) are actually communicating useful information to the other party. I might even say that omitting those interruptions (and replacing them with, say, dead silence) might be poor communication.

I really enjoyed the machine&#x27;s &quot;uhs&quot; and &quot;uhms&quot; in the demo speech. However, I felt the &quot;uh-huh&quot;s sounded forced. It&#x27;s funny how these subtleties are very important in human conversation.

&gt; as an intentional deficiency is an interesting design decision.well, in semantics&#x2F;pragmatics these discourse particles are often not deficiencies at all. They are signals with practical semantic purpose. &quot;hmms&quot; and &quot;uhs&quot; can signal attentiveness, turn-taking (turn holding, turn yielding, etc), agreement - just to name a few.For any machine system to be able to pass as human, it will have to be able to control these nuances or people will pick up on something being wrong, though they might not be able to articulate precisely what.

Ums and other filler words are not as bad as they are made out to be among the public speaking crowd: <a href="https:&#x2F;&#x2F;www.eab.com&#x2F;daily-briefing&#x2F;2016&#x2F;07&#x2F;29&#x2F;um-filler-words-may-not-always-be-bad-and-may-do-some-good" rel="nofollow">https:&#x2F;&#x2F;www.eab.com&#x2F;daily-briefing&#x2F;2016&#x2F;07&#x2F;29&#x2F;um-filler-word...</a>

People are more willing get on board with bots acting like people than the other way around.The speech disfluencies used by Duplex in the salon and restaurant interactions are perfect examples of why natural speech sounds natural. It&#x27;s the cadence as well as the timing.

Your argument about calling it &quot;noon&quot; instead of &quot;PM&quot; is just illogical. I&#x27;d always use PM instead of noon -- whenever I&#x27;m trying to be specific about something (appointments and such). I understand the argument you&#x27;re trying to make but that wasn&#x27;t enough.

If you are building a system that mimics human speech you need to teach it to be imperfect and use common parlance. Otherwise you will fall into the uncanny valley. If you listen to the conversation again there are several points where they lose immersion. For example no one would say 12 pm, they would say &#x27;noon&#x27; instead. Google has clearly done some impressive work here, and I&#x27;m now a bit more confident&#x2F;scared that they will be able to successfully fool me in the next few years.

In my city I can recognize that google has two different &#x27;voices&#x27; or voice libraries. They sound slightly different. I&#x27;m curious how that works and why it&#x27;s not all done with one.

The Brazilian Portuguese voice is not that great.

Google maps has the most pleasant and human sounding voice approach I’ve encountered in any such system.All other gps guidance voices sound incredibly crude and mechanical in comparison.

Similar to how when designing a virtual face, it looks more natural if it has some slight asymmetries and &quot;defects&quot;, and when designing a synthetic drumbeat, if it&#x27;s &quot;perfect&quot; it sounds totally robotic.Imperfection is natural and comfortable. Perfect corners and edges are artificial and weird to the distracting point.

Reminds me of comfort noise in the telephone system.Even though the system encodes silences noise free (so improve compression), it deliberately inserts noise because otherwise people think the line is dead.

Similar to lens flares in (first person) video games.

&gt; The system also sounds more natural thanks to the incorporation of speech disfluencies (e.g. “hmm”s and “uh”s). These are added when combining widely differing sound units in the concatenative TTS or adding synthetic waits, which allows the system to signal in a natural way that it is still processing. (This is what people often do when they are gathering their thoughts.) In user studies, we found that conversations using these disfluencies sound more familiar and natural.This part stuck out to me during the Google I&#x2F;O demo, as an intentional deficiency is an interesting design decision.

Currently I&#x27;m working on a similar project. Building a GAN (generative adversarial network) for voice is a lot of work and testing, you need a big dataset for labeled voices. Wavenet currently support English[0]. Also you need a neural NLP model for this language.[0]<a href="https:&#x2F;&#x2F;cloud.google.com&#x2F;text-to-speech&#x2F;docs&#x2F;voices" rel="nofollow">https:&#x2F;&#x2F;cloud.google.com&#x2F;text-to-speech&#x2F;docs&#x2F;voices</a>

I would imagine that a lot of the work that goes into making the speaking so natural would be very specific to English. As in, they would need to repeat 50% of the work for each language.

Scots are definitely native english speakers, reluctant ones at that, but english nonetheless

Do Scots count as native English speakers? What about the Irish? English has a very large number of accents and dialects that all count as &#x27;English&#x27;, but typically speech recognition software only works one variety.

In the case where the person on the other end isn&#x27;t a native english speaker, (the calling a restaurant clip) why doesn&#x27;t Google figure out what language they speak and speak it to them?

I agree, but the social pressure not to use it would be immense. Imagine the awkward moments where we start confronting people who call us and asking them if they&#x27;re real.

This reminds me of some of Isaac Asimov&#x27;s short stories in which the Three Laws are used to force robots to identify themselves.

There&#x27;s no way I should be bound by terms of service that are secret unless I know the &quot;code-phrase&quot; with which to ask for it. Google&#x27;s whole point was to deceive the user into thinking it isn&#x27;t a bot -- so you wouldn&#x27;t even know to ask it.People will be forced to &quot;discover&quot; that they&#x27;ve been lied to, when the bot is caught in a loop and they&#x27;re going through emotional distress thinking the &quot;person&quot; on the other end is having some sort of mental breakdown.

Recording calls without both parties&#x27; knowledge can be illegal.

I think at the very least the bot should disclose that it is a bot right from the beginning.

At the very least, there should be an requirement to identify when challenged.&quot;Are you a bot?&quot; &quot;Yes, I am Google Duplex v1.2. This call is being recorded; you can see our privacy policy and terms of service at <a href="http:&#x2F;&#x2F;google.com&#x2F;duplex.&quot;" rel="nofollow">http:&#x2F;&#x2F;google.com&#x2F;duplex.&quot;</a>

I am far more afraid of technology we can&#x27;t control that works for an ad company, not for us, than moving technology forward a little bit slower.We need to be able to trust our communication, and we&#x27;re moving away from that at light speed. Where a company whose entire business model is to subtly influence you in the directions profitable to them is the middleman in all of our communications.I am terrified of this. I am not sure I&#x27;ve ever seen a presentation that terrified me more that wasn&#x27;t out of a science fiction movie.

Yeah and I had your view before the last 2 years and the 2016 election. All we&#x27;ve seen is this kind of technology will be abused to the gills to create fake propaganda, robocalls, and phishing scams.I too want to absolutely, without a doubt, regulate or slow this stuff down until we can better understand how it can be abused. And WHEN it is abused, we have ways to address it and aren&#x27;t constantly playing catchup to how people are badly using these things.

I&#x27;m concerned that a lack of understanding and fear will limit the benefits we get from technology. I don&#x27;t think that calling for regulation and making this illegal is the appropriate next step.They explicitly say:&gt; We want to be clear about the intent of the call so businesses understand the context.You note:&gt; laws against what they&#x27;re doing needs to be passed.What makes you say that robots making phone calls should be illegal? This seems an odd position to have. Do you also believe that it should be illegal for robots to clean floors? Or for your machine learning spam filter to process and filter your emails?If I sound a little frustrated, I am. I&#x27;m scared that the future where technology allows us to do more with less will be blunted due to calls for making robot tasks illegal. I&#x27;d love if you&#x27;d help me understand what makes you take your position - I&#x27;m sure it&#x27;s logical, and it&#x27;s just a matter of me being able to understand your perspective. Can you help me understand?

Does your secretary or father pretend to be you when making the appointment? Google Duplex is pretending to be a human without the knowledge or permission of the person on the other end of the call.

Is it illegal for my secretary or father to make an appointment for me?

To be perfectly honest, I&#x27;ve always personally been surprised at what business people are willing to transact over the low-bandwidth media of the telephone. I&#x27;ve heard plenty of stories of people pretending to be celebrities and getting away with stuff on the phone; I don&#x27;t need a sophisticated computer to do that for me.If a real problem of deep-fakes in phone conversations emerges, people will just raise the bar of authentication in phone communications. There&#x27;s nothing magical about telephone that makes it exempt from the general need to protect against social engineering.

The reality changes. Fast. You won&#x27;t be able to trust photos, audio recordings, or videos. You will be able to trust digital signature. To have any opinion whether information about somebody else is real you will need a working web of trust. It&#x27;s gonna be a great thing and I can&#x27;t wait for it. It&#x27;s going to change news, politics and in general the world we live in. It&#x27;s going to be bigger than the Internet but way more polarizing and controversial. IMHOOne issue though, unless we hurry up and make this WoT decentralized and with open protocols, well, we will get it from FB and the likes.

How would you handle deepfakes? This is much more disturbing than what is in the post, and it isn&#x27;t a corporation who built it. I don&#x27;t think this has anything to do with regulating google.

Would you say the same thing if this tech was created by a small company?Voice copying is nothing new, many methods exists so I don&#x27;t see how Google doing this is somehow bad. It&#x27;s uncanny to say the least, but laws against tech progress? Come on :)Did you think that advancing AI will be not creepy especially when it&#x27;s good?

This is awesome, awesome tech. I just showed it to a bunch of people in my office.But at the same time, the last two years have shown us that a lot of technology is being very effectively abused to misinform at a geopolitical level, and we as a society need to better understand or regulate this stuff for sure. Lives are at stake.That said, I&#x27;d never say that Google shouldn&#x27;t work on this. It&#x27;s amazing. We just need to better understand not just how it can be used, but also how it can be abused.

Come on now, this is cool! the correct initial reaction is to be impressed.What is concerning is the surprising prevalence of technophobic notions on tech focused boards such as this.There is something perverse about wanting to punish a company for creating something cool, and it’s certainly not the way our society or politics should function.

In the short&#x2F;immediate term, it should probably be illegal to conduct a robocall without explicit opt-in of the destination caller. (Currently, if you get a non-telemarketed robocall, it is solicited, such as you scheduled an appointment and they are reminding you of it.)Furthermore, if a robocall is permitted to be made to an unsolicited destination, a robocaller must clearly identify itself, presumably starting a declaration that it is a bot&#x2F;automated agent from a given company on behalf of another company or individual.And if the call is recorded in any way, shape, or form, the terms of service would need to be presented to the callee, giving them the chance to accept or deny said terms. (Note that if the example calls at I&#x2F;O were not staged, and real calls, I suspect the recording of them would be illegal in many jurisdictions, including California.)

Given the two technologies you&#x27;ve highlighted, what specific regulation would you recommend be added to prevent what bad behavior?

The irony of someone named ‘...trekkie’, a fan of a future where a military alliance of uber powerful weaponry capable of destroying populations, while continually proclaiming they’re on a mission of peace, featuring computers pretending to be people and doing work for them, androids, and synthetic holographic ar&#x2F;vr people criticizing primitive AI tech is a bit rich.We’ve arrived at full Luddism now where life saving and time saving technology in fields like health, transportation, and customer service will be inhibited by hyperbolic fearmongering.Are you going to pass laws against realistic sounding synthetic voices? Against computers that understand queries “too well”? Against self driving vehicles that drive better and safer than people? All on theoretical harm that actually hasn’t taken place because you’ve watched too many dystopian Netflix sci-fi episodes?I don’t know what’s more dangerous, the real Skynet, or people who might harm millions by voting for political policies that inhibit real improvements that could be made to help them.At least, if you want to talk AI being used for harmful things we could discuss feed optimization that parasitizes people’s attention spans and keeps them glued or wasting money on pulling more slot machine levers. Technology that wastes people’s time and money as opposed to things that make people more efficient.

Never have we had a more clear example of why Google needs to be regulated, and laws against what they&#x27;re doing needs to be passed.The fact that they think it&#x27;s okay to have bots pretending to be human making phone calls, shortly after demonstrating how quickly they can copy someone&#x27;s voice (re: John Legend), it shows a blatant disregard for what they&#x27;re creating.

The technology promises &quot;wealth and leisure for all&quot;. The capital owners promise &quot;it&#x27;ll trickle down&quot;. Technical utopianists need to start tempering their optimism with the realities of human nature and design systems accordingly.

&gt; I&#x27;m very excited for a future where more and more tasks can be automated, enabling humans to get higher and higher living standards.That&#x27;s been the dream for a long time in some circles. With the enormous productivity gains and ability to leverage external energy sources (fossil fuels, solar, etc.) we could have built a society of wealth and leisure for all.Maybe we will still. The hope is that if there is enough of a productivity gain in a short enough time period (like the introduction of AGI powered robots) that this could still happen.

I recommend you look at the statistical evidence. The average wage in the developing world has doubled, and the global poverty rate has halved, over the last 20 years. That&#x27;s all thanks to automation.

What past productivity improvements and&#x2F;or economic developments have ever pointed to your fantasy being realized? How will thousands of workers in the Philippines and India losing their call center jobs result in &quot;higher and higher living standards?&quot;I&#x27;m sorry but given historical context and popular capitalist intent, this future you speak of is simply fantastical. The working class would sooner be made extinct before a Jetsons-esque future of leisure for all came to fruition.

Alright, let&#x27;s do a small breakdown:20% of all jobs will be lost to self-driving cars25% of all jobs will be lost to automated customer support&#x2F;marketing&#x2F;non-direct human interaction10% of all jobs will be lost to automating away all middlemen5% of all jobs will be lost to automating away rudimentary programmingNow, who is going to win here?

This is amazing! I&#x27;m very excited for a future where more and more tasks can be automated, enabling humans to get higher and higher living standards with the same&#x2F;fewer input resources.In short, technology is a wonderful thing that allows very low marginal costs. This is what we need to make the future a better place, given a consistent or growing population.&quot;Technology is miraculous because it allows us to do more with less.&quot; This is a perfect demonstration of that.See also: <a href="https:&#x2F;&#x2F;www.youtube.com&#x2F;watch?v=rvskMHn0sqQ" rel="nofollow">https:&#x2F;&#x2F;www.youtube.com&#x2F;watch?v=rvskMHn0sqQ</a>&quot;A Selfish Argument for Making the World a Better Place – Egoistic Altruism&quot;

Wish they can provide the integrated package as an API.

I wonder if they detect that and solve whatever task it is directly then.

The most inefficient API you could design!

Or Alexa answers the phone and refuses to speak to it.

I wonder if it would be able to tell if it was talking to another Duplex bot and instead of speaking in English, it would communicate more efficiently.

Just wait until a Google Duplex caller &quot;on behalf of a client&quot; calls to schedule a reservation at a restaurant using Google Duplex to answer the phones.

Can&#x27;t decide whether this is amazing or scary.

I don&#x27;t think the scaling concern should be underestimated. Thats a big difference. Its mostly a numbers game.

So now I can script a bot to book restaurant reservations all over the city at busy times. Then nobody shows up for the reservations, the busy time has passed, and customers have moved on or gone home.Restaurants make or break on one or two nights in a month. A calculated social engineering attack like this could bring down hundreds of restaurants in a city, which would cause millions of dollars in lost taxes, and you see where this is going.

At least in the demo, the system didn&#x27;t identify itself as a bot. It pretended (fairly convincingly) to just be a person, and seemed focused on interactions (making a restaurant reservation, etc.) that don&#x27;t involve someone recognizing your voice. In other words, I don&#x27;t think the attack surface is any different than situations where you could just place the call yourself, save perhaps for the possibility of scaling it.

Or maybe we should fix the phone system. Something is terribly broken when my connection to YouTube is better protected and authenticated than when I&#x27;m talking to the IRS over the telephone.

This isn&#x27;t that different than all the bad actors calling from the &quot;IRS&quot;.  Educating people about giving personal details out over the phone has always been a challenge and will continue to be one.

I don&#x27;t understand what you mean, can you elaborate? Who calls the 1-800 number, and what info do they get from the call?

Credit card companies have already solved this problem by publishing a well-known 1-800 number that their customers can call back to verify the intent of an unexpected caller.Google Duplex can do the same thing - without even staffing a call center.

That&#x27;s now how this works, that&#x27;s not how any of this works. If you have ever been cold called by Google I would go change your passwords and the like.

When people talk about AI taking over, they really need to consider this route. Social engineering most of the world simultaneously. It doesn&#x27;t have to build armies of robots when we&#x27;ll tear everything down for it.

Clever social engineering is really customized, but social engineering that works &quot;often enough&quot; is mostly reducible to a flow chart.If you can automate it, your response rate only needs to be, what, like people clicking on spam? Tiny.The days of trusting meat sacks with important information might be numbered.Someone should make the dystopia where Skynet doesn&#x27;t build terminators, but call centers.

On Windows, there&#x27;s a group policy to require Ctrl-Alt-Del before the UAC prompt asks for your password. It&#x27;s impossible for an application to hijack that key combination.

So you simply tell google a cookie that it would keep safe, to prove it’s google when calling. Presumably it wouldn’t be on speakerphone at the time.Have you ever wondered the same thing about your favorite OS’s admin credentials dialog? What if an app spoofs it?

Click all the boxes in this picture that contain a sign.

Has anybody found a way&#x2F;hack to bypass Recaptcha ?

Sounds like we&#x27;ll have to start solving voice captchas just to call businesses.

I think the biggest issue won&#x27;t come from phishers, but from resellers and other occupations that are helped by automation. I could easily call every hot restaurant in San Francisco a several a day to try to snatch a table.On the other hand, most of these places have online reservations which are already extremely gamed.

And without some indicator of attribution, this is just doubling down on the problem of Caller ID spoofing.<a href="https:&#x2F;&#x2F;www.nytimes.com&#x2F;2018&#x2F;05&#x2F;06&#x2F;your-money&#x2F;robocalls-rise-illegal.html" rel="nofollow">https:&#x2F;&#x2F;www.nytimes.com&#x2F;2018&#x2F;05&#x2F;06&#x2F;your-money&#x2F;robocalls-rise...</a>

So what can you trust? The person calling you might be a simulated voice, crafted to sound like someone you know. The person texting&#x2F;DMing&#x2F;emailing you might be an attacker pretended to be someone you know. You can put trust in things like PGP, but they&#x27;re sparsely adopted and still leave a huge attack surface.

Either that, or it will train people to stop trusting a phone call as some sort of authentication.  The whole social engineering problem is that when somebody gets a phone call, they trust that the person at the other end is who they say they are.  Which can be a false assumption.  When the robots start calling, maybe people will finally stop making that assumption automatically.

I don&#x27;t think the intention of this system is for your boss to use Google to call you and ask for passwords. It seems very limited to scheduling appointments.

I stopped answering calls I am not expecting years ago. If it&#x27;s important they will find a better async way to get in touch.

This is going to be awesome&#x2F;terrible for social engineering attacks. We&#x27;ll be training millions of people to to trust phone calls from &quot;Google&quot;. But are you talking to Google-calling-on-behalf-of-your-boss, or Google-calling-on-behalf-of-some-phisher? Or better yet, some custom system that pretends to be Google over the phone? Who knows!

I think they meant: are you really talking to your friend the human, and not a bot impersonating that friend.

If you can&#x27;t tell if your AI friend is real, are they not a real friend? They might not be human, but they will still be your friend, no?

It&#x27;s already happening, most people I know rarely, if ever, answer calls from unknown phone numbers.

And claw at your own eyes to make sure you are not in a VR :)

But then you would need to shoot them to see if they bleed or not.

Nah because with everyone using AR, Google Glass, and Snapchat spectacles, we won&#x27;t even be able to trust what we&#x27;re seeing isn&#x27;t manipulated.

One positive thing about our dystopian future is that it will force us to meet up more in person, if only to confirm our friends are real.

I just hope it never gets one of the 9 out of 10 humans who don&#x27;t repeat back important information to prevent miscommunication...

Should I be surprised that it happened to work in the example they decided to showcase in the release announcement?I&#x27;m afraid I foresee this whole thing going quite hilariously wrong on the order of magnitude of Microsoft&#x27;s Tay <a href="https:&#x2F;&#x2F;en.wikipedia.org&#x2F;wiki&#x2F;Tay_(bot)" rel="nofollow">https:&#x2F;&#x2F;en.wikipedia.org&#x2F;wiki&#x2F;Tay_(bot)</a>

Listen to the audio examples. The person at the restaurant misheard the bot and repeated incorrect information back to it and it handled it in stride.

There is something so wrong with this comment that all the problems with today&#x27;s tech culture can be extrapolated from it...

&gt; I&#x27;d expect to tell it to book a 6pm table for 2 and wind up with an 8pm table for 10.That&#x27;s a win, consider: It&#x27;s the right restaurant, on the right date and 10 is larger than 2 so there&#x27;ll be room enough. Ok, you have to wait two hours but given that this tiny place is normally so busy you have to book in advance what you lost in time waiting you will more than make up by the fact that you&#x27;ll be the only people there expecting service.

I&#x27;m assuming, perhaps incorrectly, that Google would have one backend service or set of services for all speech to text translation, and that all related products would be powered by it. They&#x27;ve been iterating on speech to text interfaces for at least 15 years now.

You&#x27;re comparing different generations of technologies.

I&#x27;m so very disappointed by google home. It feels like they left this project behind. It&#x27;s infuriating how dumb this machine is, and how many times I have to repeat &quot;ok google, hey google&quot; for my old google home device that was working way better at the beginning. Google assistant on android also seems to work much better.

I&#x27;m with you. We&#x27;ve seen a demo, not a real product that people are using. I wish people would temper their expectations a bit. If product development has taught me anything, it&#x27;s that it&#x27;s easy to put together a demo to show how well a product can work in the golden path, but once you stray from it, it becomes incredibly difficult to solve all edge cases. With something like conversation, there are lots of edge cases that need to be handled. Once you have a missed edge case, the illusion of reliability falls apart.

The Pixelbuds were supposed to be something similar, but did not live up to the hype....and I&#x27;ve experienced the same things with my home automation with Google Home, and don&#x27;t get me started on using it so control Spotify etc..

Exactly my thought!  I appreciate Google&#x27;s effort in many fields such as distributed computing but in the past few years, a lot of Google IO announcements feel like vapourware&#x2F;demoware that they&#x27;re going to scrap&#x2F;rebuild in a few years.

Google home&#x27;s speech recognition can&#x27;t even reliably turn my lights on and off every time, why would I trust it to book a restaurant reservation? I&#x27;d expect to tell it to book a 6pm table for 2 and wind up with an 8pm table for 10.

&gt; Ironically they may add that at some point, and Duplex just calls on my behalf without me even knowing.Yup, good observation; seems highly likely, save for the &#x27;without even knowing&#x27; bit... I would imagine you&#x27;d &#x27;request to book&#x27; and an async Duplex operation would run in the background and send you a notification of the outcome &#x2F; possibilities.

I can imagine this being an extension to robots.txt - i.e. do I mind my business being robo-called, or would I rather send an API call to &lt;x&gt;. This is an interesting point in AI technology.

There are plenty of places in my town that don&#x27;t have online ordering or booking...this would (potentially) help with that...but really, I&#x27;d rather there be more tech be built around enabling theses companies to &quot;be online&quot; such as an online ordering or booking system. When I find a hair cut place, why can&#x27;t I just book an appointment right from the Google Maps search? I&#x27;d rather have that type of convenience instead. Ironically they may add that at some point, and Duplex just calls on my behalf without me even knowing.

I wonder who’s have the last laugh on this.

Google voice API, like a real API but much slower.

&gt; And then the salon and the restaurant start using Duplex as well, and all phone conversations become Google talking to Google.This must be the most convoluted API protocol ever invented.

And then the salon and the restaurant start using Duplex as well, and all phone conversations become Google talking to Google.How bout: &quot;Hey Duplex, call this support number and get a top level human manager on the line please.&quot;Then we get a HN article: &quot;Duplex is fighting Alexa!&quot;

The examples they give are quite impressive but I always want to hear some examples of failures as well with these types of technology. They mention that the system is self monitoring and will try to detect a situation it can&#x27;t handle and redirect to a human operator but I think some examples of situations it can&#x27;t handle or where it gets things wrong would be very useful in understanding its real world robustness.

Stuff like this makes me excited for the world my kids will live in.

Exactly this. I put together a rough demo of similar AI scheduling in the past and x.ai does this as well. The AI works within a given constraint before proposing new times because the idea is to free up time for you, not make you clean up a scheduling mistake every time something is decided by the AI.

I&#x27;d expect you to be able to give Duplex the range of dates and times where you&#x27;re available (or let it get the info from your Google Calendar).

Ideally this would have to have access to your calendar or be given a time range of options.

This is awesome, ButThe algorithm of cause doesn&#x27;t understand all contexts. What troubles me is that, in the first example, should we really give algorithm the freedom to propose a new date for an appointment? It reminds me one thing that particular bothers me with the Gmail&#x27;s smart reply feature, where when given Monday or Wednesday as options, the suggested reply is, &#x27;How about Tuesday&#x27;, which does make the conversation flows, but doesn&#x27;t really make any logical sense.It makes a good demo, I am very much impressed, however, I feel it will run into a LOT of issues, even only in those provided scenarios, should those scenarios become more sophisticated.

People have already had good success engaging telemarketers with much more primitive robots. Here are some funny samples on YouTube:Lenny:
<a href="https:&#x2F;&#x2F;www.youtube.com&#x2F;watch?v=LgT44DuIaAM&amp;list=PLduL71_GKzHHk4hLga0nOGWrXlhl-i_3g" rel="nofollow">https:&#x2F;&#x2F;www.youtube.com&#x2F;watch?v=LgT44DuIaAM&amp;list=PLduL71_GKz...</a>Jolly Roger:
<a href="https:&#x2F;&#x2F;www.youtube.com&#x2F;channel&#x2F;UC3OxCWLEmoIhNMm-hnvBm9Q" rel="nofollow">https:&#x2F;&#x2F;www.youtube.com&#x2F;channel&#x2F;UC3OxCWLEmoIhNMm-hnvBm9Q</a>

Neural networks are a fundamentally statistical technique, so you find the state of the art whereever you find the largest scale. No one is operating at larger scale than Google, which has the second order effect of attracting the best talent, which then gets sorted to the highest priority problems. There are a non-trivial number of instances where Google will beat their own state of the art, and I&#x27;m quite confident they are sitting on further results to avoid a public embaressment of riches.

Can someone from the research community comment on how this compares to the state of the art in open research?

Crazy! Looks similar to Upcall.com where they have real people calling for you.

At last, putting the power of annoying robot phone trees in the hands of the consumer, to be directed at businesses.For that reason alone, I&#x27;m excited. ;)(Note: In fairness, the conversation demos are actually really slick and much better than a phone tree. I&#x27;ll be interested to see how well it works in practice.)

You are probably right that in principle one could eventually come up with a full catalog of features of a reservation. There would be about, say, 100 of those.I seriously doubt that they will proceed to define and collect them, since those are probably 10% or less of all reservations, but lets say they would.Then still, the conversation you make to make the reservation is a process in which you make the decision.Say, there is a place inside at 20:00 or a place in the garden at 20:30. Are you going to let Google choose between the two options for you?Do you imagine there would be an api in which you specify to the assistant, before it makes a call, your preferences in that much granularity?

The bot can gather this restaurant specific information over several conversations with the restaurant. This wasn&#x27;t possible before. This domain isn&#x27;t too wide.

They have listed the problems with such systems in the first paragraph. They claim to overcome these by restricting to very specific domains. But specific domains are usually still wide. A table for two, but in the garden or inside? Inside on which floor? Sometimes it doesn&#x27;t matter, sometimes it does, and these things are specific and different for every business.Really skeptical about this. And if this does become a thing, it will dumb down the interaction.

Not that I&#x27;ve seen. The most advanced IVR systems that some banks are using have voice fingerprints and recognition etc but the TTS is still noticeably robotic.Other companies are using huge libraries of recorded human voice for communications and concatenating them together in intelligent ways.

Is any product remotely as close to this with natural-sounding human speech? Are we entering an era where all voice assistants are getting to that &quot;Her&quot; level in terms of human-like vocal quality?That part is as impressive to me as the semantic parsing it&#x27;s doing on that call.

Well, I&#x27;m already receiving automated calls about &quot;that car accident&quot; and they connect me to a real person if I happen to respond with a predefined keyword.I would say that there&#x27;s a market for that. How long before rough agents implement good AI for their operations? How long before we implement defences against AI?

And you think this is the end of innovation, this is it, no improvements from here?Duplex is interacting with a human and that person has no idea its a computer on the other end. Yes, Duplex is limited as it stands, but what is there to make you think something as described in grandparent post won&#x27;t exist in ten years?

&gt; Yes, duplex is for completing specific tasks, but how much of a difference is there between &quot;Duplex, book a table for four at 8pm&quot; and &quot;Duplex, ask my victim about how their day was&quot;?The former (booking a table) is much more &quot;constrained&quot;, as in the conversation would most likely not go into much of a tangent, because there are only so many responses to a statement like &quot;book a table for four at 8pm&quot; (8pm is full, 8pm works, etc).Whereas asking someone how their day was would give the &quot;victim&quot; a much bigger breadth of responses (and additional questions!) that would cause the AI to stumble and fail to give a satisfactory answer. That, and running this 1,000 times simultaneously so that no one person would be able to intervene to &quot;help&quot; the AI would just be a highly unscalable operation.So yes, huge difference.

Duplex problem is not specific tasks but limited domain. If you are engaging with a person, you are going to err on multiple domains randomly. Especially if you are building a rapport over a long period of time.If an AI can do that, then singularity is definitively reached.

This is one end-game of AI - changing the economics of scams.Imagine how a &#x27;long con&#x27; works today: a scammer befriends a person online through a video game or social platform, and develops a rapport with them over the span of days, weeks, months. After some trust has been gained, the scammer then requests money from the victim. Does this happen a lot today? I don&#x27;t know, but certainly one reason it doesn&#x27;t is the economics of the scam. Who wants to spend a significant period of time gaining trust just for the chance of a payout?AI is going to flip this on its head. Rather than dedicating hundreds of hours of a scammers time, a scammer could instead use a system like Duplex to befriend hundreds or thousands of victims simultaneously. Let it run for a few months, developing a strong rapport with the target, until the AI finally requests some money from the victim.Yes, duplex is for completing specific tasks, but how much of a difference is there between &quot;Duplex, book a table for four at 8pm&quot; and &quot;Duplex, ask my victim about how their day was&quot;?

This is truly amazing work! Congrats Google for achieving this :)

Give your credit card number to one of them and it will be even more fun.

If the main impediment to dining at a restaurant was having to reserve a table, I think that would be a real possibility. But the main impediments are usually cost, complexity in getting to the restaurant, unfamiliarity with how good the food and service are, agreeing with the rest of the party about the timing, etc.. The actual booking of a table is a minor stepping stone on the road from the thought to the actual execution of the plan.

I worry that this will make people have an AI make phone calls for them regularly when they wouldn&#x27;t have in the past. At some point, the restaurants will need to have their own AI to respond to the call volume. You end up with 2 AIs attempting to have a conversation.

&gt; The people losing their marbles over this being some kind of Turing Test passing distopian stuff are missing the point at how limited this domain is.Right, and this kind of comments will continue for a while.The question is - when the &quot;this is trivial, move on&quot; type of comments will start to fade out? Five years? Ten?

Ya, I agree that an API is the right solution but the benefit of this is that both sides aren&#x27;t forced to adopt it at the same time, it&#x27;s more resilient to changes on either side...

Where did I say it’s trivial? What I’m saying is, people ar anthropomorphizing and extrapolating what this system is doing far above what it is actually doing, and then using it to justify fears that Skynet’s around the corner.This system can’t pass the Turing test, it would be fooled probably by a simple question about itself or a subject outside the domain, like the kind of food you like.You’ve got people in this thread hyperventilating about AI duping your voice and the. Becoming a doppelgänger and therefore we need laws immediately to stop this dystopia? Let’s calm your cortisol levels for a second and stop acting like Thanos just got the last gem.

&gt; The end game is clearly to use an api, not this.My understanding is API integration is what wechat is in china -- every hair salon and equivalent-of-corner-pizza shop has some wechat integration, payment and all.Voice bots like this will have the advantage of ubiquity. At least a couple years ago before every resteraunt had 5 tablets for all their seamless&#x2F;grubhub&#x2F;chowhound&#x2F;whatever apps, pretty much the only reason the fax machine was still around was for restaurant ordering. Although there were clearly better ways of doing it (see how Dominos reinvented itself as a tech company), the sheer ubiquity of fax as the lowest common denominator kept the tech around.In that light, it&#x27;s kinda like the cell-phones-leapfrogging-landlines-in-developing-countries argument... part of the wechat story involves a massive population entering the consumer class at a time when everything was digital. Call me out if this is a gross over-generalization, but in a way, the wechat population never had to deal with the backwards-compatibility of people growing up ordering a pizza over the phone.It&#x27;ll be interesting to see how the API-centric approach (wechat) plays out versus the lowest-common-denominator ubiquity approach (voicebots). I&#x27;d stop short of calling API&#x27;s the end game though.

The benefit is that a human could jump in and take over either side of the conversation.Anyone can have a conversation, but not everyone can author an API.

Please go and convince the mom and pop hardware store or bakery in your town to use an API instead of answering the phone.

If you ever needed proof of the negativity in the tech community, this thread is it.Google actually delivers a real world product incorporating the most advanced AI we&#x27;ve had the chance to experience, and half the HackerNews comments are &quot;Wow, this is so dumb, can&#x27;t wait for it to become technical debt in 10 years&quot;.

I love how no matter how amazing something is, someone will eventually say it&#x27;s trivial, even though it&#x27;s taken the smartest people on earth decades to figure out how to do this.&gt;The end game would be for the business to run something like duplex on the other side, and you’d have duplex talking to duplex.The end game is clearly to use an api, not this.

You’re totally right.However while this is useful to bootstrap a new technology rollout, 10 years on its just technical debt.The amount of tech debt in the system behind credit cards is crazy, because originally charges where phoned in to the card issuer manually, and everything from then on - magstripe, chip &amp; PIN, online only transactions, etc, has all been built on top, and the leaky abstractions show through in daily difficulties with the card system for end users, like lack of real-time balance (in some cases), lack of transaction metadata, etc.

Doesn&#x27;t most carriers heavily &quot;compress&quot; the sound, removing all sounds&#x2F;frequencies that a human can&#x27;t hear, etc? <a href="https:&#x2F;&#x2F;www.youtube.com&#x2F;watch?v=w2A8q3XIhu0" rel="nofollow">https:&#x2F;&#x2F;www.youtube.com&#x2F;watch?v=w2A8q3XIhu0</a>

I like this. So basically the old school modem sound, but in frequency that can&#x27;t be heard.  It would only take a fraction of a second to send out the feeler, and would not be noticed if a live human picked up. Could even detect a human and send the call over to a live representative without anyone noticing.

Yes, it would be nice if in parallel Google came up with open machine-friendly protocols for each of the use-cases Duplex supported, with a clear migration path away (e.x. businesses started publishing the endpoints and protocols they supported alongside their phone number so you could skip the call completely)

Instead of a high frequency tone, just watermark the background noise or the speech pattern. You could watermark the background static, the voice samples, or even the speech patterns. All you really need is something like 30 bits of data to identify a call as a Duplex call with very high probability, and I’m certain you can find a way to imprint that many bits into the frequency spectrum of your background noise.

&quot;Ba weep granna weep ninny bong&quot;It is the universal greeting for cybernetic organisms, after all.

maybe just cut the chase and do an API for reservation?

Why not just a pattern of umms and arrs that it already seems to add into output. Easier to detect for it and harder for a human to recognise.

To save CPU&#x2F;GPU&#x2F;TPU there should be a high-frequency sound, as in people can’t hear, so the computers talking to each other and switch to a faster way to communicate.
If this is included you also have way to detect if you are talking to a bot&#x2F;duplex.

They&#x27;re not unaware of this concern. From the article:&gt; The Google Duplex technology is built to sound natural, to make the conversation experience comfortable. It’s important to us that users and businesses have a good experience with this service, and transparency is a key part of that. We want to be clear about the intent of the call so businesses understand the context. We’ll be experimenting with the right approach over the coming months.

Not sure why you&#x27;re using future tense here.

If you can&#x27;t tell whether you&#x27;re talking to a computer, does it matter?

If we accept your premise that bots become so close to humans that they are virtually indistinguishable, then at that point does it matter who the &quot;person&quot; on the other end of the line is? I&#x27;d argue it doesn&#x27;t because the outcome would be the same.

Phoning like this you already have to wonder if the person on the end is an idiot who will screw stuff up. I&#x27;m not sure it maybe being a google bot will make much difference.

What happens then if I&#x27;m a human that masquerades as a computer? Seems like a neat way of explaining away a number of social faux pas or drunk-dialling. &quot;Me? Noooo... that must have been my PhoneBot 2000. Supposedly the next firmware update should solve that kind of problem.&quot;That said, I don&#x27;t necessarily disagree; there is going to need to be lots of these kind of issues that need sorting out before we reach a Culture-level of AI interaction.

The problem is that computers will now be interacting with people and people will become unsure of whether they are taking to a computer or a person.  It will create a bewildering world full of mistrust. I would argue that there should be a law proclaiming that computers must identify themselves.

The point being is that it&#x27;s a human speech API which is easier for humans to utilize &quot;manually&quot; if they don&#x27;t have a robot servant.

Yes!This is an API that requires no computer on the user&#x27;s end and is portable across different implementations from different companies.It&#x27;s not ideal. Actual standardized APIs are better. But, uh, have you ever worked with industry standard APIs? I have, and standardized is not how I would describe them.

This is my initial reaction too, but APIs need work for coding, integration, testing, make sure data is sent in the right format, etc, while voice-robot-to-voice-robot will just work out of the box.

It&#x27;s not just about the user. There&#x27;s no open, universally agreed upon booking system API...except for human voice.This is creating a natural-language based booking API that any system or business can tap into

&quot;Duplex seems trained against this corpus. The end game would be for the business to run something like duplex on the other side, and you’d have duplex talking to duplex.&quot;So, you would just have an automated booking system API which is better handled by not placing calls as its form of communication. Right?

Maybe a general AI is just a very big collection of many smaller abilities.

Absolutely correct. And so-called general AI may never happen. Regardless, this is shocking. It immediately needs to be factored into any speculation about what the world will look like in 20 years. Innumerable questions.

&gt; But that in itself is not even true across the industry, some(most) phone bookings are very complexCitation needed for that &quot;(most)&quot;. I work for a company with a call center and a large part of calls are simple ones that could be easily answered by just reading the FAQ page on our website.&gt; otherwise they would just use a web interface.I think the problem is more about resources. My local hairdresser use his phone and a notebook to take bookings. It takes a bit of his time and could easily be replaced by a Web interface but he doesn’t have any resource for that (and some people still prefer using their phones).

&gt; People who answer phones to take bookings perform an extremely limited set of questions and responses,But that in itself is not even true across the industry, some(most) phone bookings are very complex, otherwise they would just use a web interface.

Likely it&#x27;ll be a transition technology for businesses still using phone appointments as their primary API, and eventually it&#x27;ll ease the path for more direct integration.

Because businesses can offer Duplex for customers to call and speak to after hours if they still want to keep a human phone assistant. It&#x27;s modular, not Rube Goldbergian.

You also seem to completely miss that outside of silicon valley, large portions of the population actually want to interact with things by phone.That&#x27;s changing for sure, but the demand for phone based services is still very high.

Because there are lots of businesses that don&#x27;t have a booking API and don&#x27;t see the need for one, or can&#x27;t afford one. This kind of technology allows interaction with them, because it&#x27;s easier to interact on a common transport protocol than to expect everyone to change to your preferred one.
That being said I feel like it won&#x27;t be long until this tech is used for scamming, phishing and pranks.

It&#x27;s specifically for the salons, yoga studios, restaurants, etc that don&#x27;t use any kind of online booking system.

In the second example they ask what&#x27;s the line likely to be like on Wednesday. Show me one API that does that. English is richer than APIs.

That end game seems very wasteful and Rube Goldbergian. Why use an 1800s technology as the transport layer when salons, yoga studios, and more already use things like MindBody, which already has an appointments API? I’d honestly be way more interested if this integrated with MindBody, OpenTable, DMV websites, car dealer appointment systems, medical office scheduling systems — all of which already have APIs or at least web pages. But then, saying that you wrote and will maintain some WWW mechanize stuff that posts forms is way less marketable to the general population who see this as magic.Also, they’ll discontinue it after a year once it gets enough negative press about how it doesn’t work well and loses business for businesses.

A large part of introducing new tech is enabling the transition from existing tech. Maybe if we scrapped all old cars and allowed only autonomously vehicles we’d be be sleeping at 120 km&#x2F;h next week. But some businesses still run on COBOL.

Lol imagine my phone talking with an automated customer service line. Two machines, talking to one another, not using any of the existing protocols. My phone would have a database of questions to ask, form it into an English sentence, run it through a text-to-speech, transmit this to the other phone. Their &quot;phone&quot; would run a speech-to-text, run NLP, match it with its own database, and do the whole thing again in the opposite direction.This gives a whole new meaning to &quot;all of UI&#x2F;UX is basically prettifying database queries&quot;.

The person who invents a way to change the little green phone icon with a little human icon for when they want to talk to a human will be a zillionaire.But that is the far future. Realistically, I just don&#x27;t see this as feasible any time soon.

This really made me laugh. You are also totally right, we have better solutions to this problem already.

This doesn&#x27;t even need to happen on the voice connection. A register of &quot;does this number map to a known system&quot; would be enough. Then it&#x27;s just up to common APIs.

There are already apps&#x2F;technologies that transmit information through audio at frequencies not audible to humans. It should be trivial to adapt this so that if two AI systems are interacting they can perform an &quot;AI-handshake&quot; in the audio at the start and then switch to a more efficient form of communication.

Lol, but if you think about it, what stops businesses from doing this today?It&#x27;s the lack of a universal API.If a barber shop wants to make it possible for a 3rd party app to book appointments then they have to release some API. But that&#x27;s not the end of it. The 3rd party app has to first discover their Api, someone has to understand it and write code to use it, and then deploy that code.This is a problem today because there is no universal Api that all services can useWith Duplex, verbals speech becomes a universal Api that every service can parse and communicate to each other wtih. Also, the discoverability is taken care of by using publicly cataloged phone numbers on services like Google Maps, Yelp, etc

...somewhere, Kevin Mitnick just smiled.I like to think it was a smile of renewed relevance due to unbelievably poor technical decisions.

Google Duplex is just a backwards compatibility shim...

That form of communication exists. It&#x27;s called REST.

The issue here is interfacing with ancient telephone systems.If the two bots were to slip in some subliminal beeps and boops to recognize each other; then they could change their speech to very quick binary communication.

you mean the entire tcp stack and network to network communication?

I can already hear the modem handshake playing in my head...

You can&#x27;t just drop compatibility. We will have A.I. trained voice systems that mimic natural speech just enough to be understood by duplex while compressing the exchange to a minimum. Data transfer will be measured in microwords per second. Future versions of duplex will of course detect this kind of compressed speech and reply in kind, falling back to normal speech only if the immediate response is similar to north american confusion.

&gt; The end game would be for the business to run something like duplex on the other side, and you’d have duplex talking to duplex.And looking even further into the future, we can imagine a day when the computers forgo natural speech and use a better-suited form of communication. Some kind of sequence of ones and zeros transmitted directly across the wire.

When you mash 0 with Comcast, it kind of scoffs at you and tells you that it needs some information before it can help you. If you keep hitting 0 it just hangs up. Pretty terrible experience, but they don’t have to care because they have monopolized so many neighborhoods.

As long as there&#x27;s a way out of the AVR system... too often, you get trapped when you want something that isn&#x27;t an option.

Yes, that first restaurant reservation would have taken me several times as long to complete.

&quot;The funny thing about AI is that it’s a moving target. In the seventies, someone might ask “what are the goals of AI?” And you might say, “Oh, we want a computer who can beat a chess master, or who can understand actual language speech, or who can search a whole database very quickly.” We do all that now, like face recognition. All these things that we thought were AI, we can do them. But once you do them, you don’t think of them as AI. It has this connotation of some mysterious magical component to it, but when you actually solve one of these problems, you don’t solve it using magic, you solve it using clever mathematics. It’s no longer magical. It becomes science, and then you don’t think of it as AI anymore. It’s amazing how you can speak into your phone and ask for the nearest Thai restaurant, and it will find it. This would have been called AI, but we don’t think about it like that anymore. So I think, almost by definition, we will never have AI because we’ll never achieve the goals of AI or cease to be caught up with it.&quot;

Dude, this system has better conversation skills than me. I mean literally. Well, I&#x27;am autistic + esl (+ kinda too why). But still it&#x27;s kinda incredible that a system is actually better.

I think it&#x27;s more unfortunate that so many people are just so opposed to picking up the phone and talking to someone.

&gt; There are services like OpenTable but even here in Silicon Valley only a small minority of restaurants use them.OpenTable takes a cut, no? That always going to limit availability.

It&#x27;s unfortunate that in 2018 we still have to resort to klugey work arounds like this with so many restaurants and hair salons instead of being able to make reservations online. There are services like OpenTable but even here in Silicon Valley only a small minority of restaurants use them. It seems like there&#x27;s a huge opportunity if someone can crack the market.

Uh...businesses have been automating phone calls for decades now

&gt; Most people working in hair salons or restaurants are very busy with customers and don’t want to handle these calls, so I think the reverse of this duplex system, a more natural voice booking system for small businesses would help the immensely free up their workers to focus on customers.I think this perspective is very short sighted. You will lose customers to automation, but businesses wont turn away customers because of automation.Customers and prospects don&#x27;t want to interact with machines, but businesses should be willing to give customers what they want.The idea that a tool can be rolled out to millions of consumers, even with limited use cases and not have to get adoption from businesses to be useful is IMHO a much bigger opportunity and much better use case than rolling out a tool to businesses that make the interaction less personal.Customers need to trust businesses, business only need to collect money from customers.I think everyone who focuses on chatbots from the business use case perspective is missing the bigger opportunity.A technology that can give a consumer access to ALL businesses, not just the ones who adopt a new technology offers much more utility than serving businesses or the shortsighted use cases like saving time and money for the business.

No, it doesn&#x27;t.The problem of universal API is entirely orthogonal to voice communications. Duplex is not a Turnig-complete system, it&#x27;s just an API behind a voice recognition layer. All the important problems for universal APIs happen after that layer.Ultimately, what you describe can work perfectly only when everyone is using Duplex, which is equivalent to everyone using Google-defined API. That&#x27;s not universal, because you have one entity behind it.The only way this brings us somewhat closer to universal API is that if you expect it to handle humans as well, it introduces some constraints to the space of possible APIs, which could make it easier for everyone to agree on a common format. Constraints of natural language processing without a human-level AI require your API to be very fuzzy and very lenient. There&#x27;s nothing stopping one from implementing those same constraints over a text or binary protocol. Nothing except no reason for businesses to do it.

As crazy as it seems at first glance, a double Duplex system would be a really beneficial result.What stops businesses from setting up Apis from scheduling services today? It&#x27;s the lack of a universal API.If a barber shop wants to make it possible for a 3rd party app to book appointments then they have to release some API. But that&#x27;s not the end of it. The 3rd party app has to first discover their Api, someone has to understand it and write code to use it, and then deploy that code.I&#x27;ll repeat for emphasis: This is mainly a problem today because there is no universal Api that all services can useWith Duplex, verbals speech becomes a universal Api that every service can parse and communicate to each other with. Also, the discoverability is taken care of by using publicly cataloged phone numbers on services like Google Maps, Yelp, etc

cromwellan didn&#x27;t say it was easy. It&#x27;s still hard to do something limited.

You gotta be kidding. If it was that easy why didn&#x27;t you build it yourself? Easy billion right there.

The people losing their marbles over this being some kind of Turing Test passing distopian stuff are missing the point at how limited this domain is.People who answer phones to take bookings perform an extremely limited set of questions and responses, that’s why they can even be replaced by dumb voice response systems in many cases.In these cases, the human being answering the phone is themselves acting like a bot following a repetitive script.Duplex seems trained against this corpus. The end game would be for the business to run something like duplex on the other side, and you’d have duplex talking to duplex.Most people working in hair salons or restaurants are very busy with customers and don’t want to handle these calls, so I think the reverse of this duplex system, a more natural voice booking system for small businesses would help the immensely free up their workers to focus on customers.

Google Duplex: An AI System for Accomplishing Real World Tasks Over the Phone