Shit, Google is gunning for my job talking on the phone

[Casual Observer]

I wasn't particularly worried about the future of jobs like I have (selling tech stuff over the phone) because AI can't really sound human, right??? So now Google Duplex can make simple calls and sounds remarkably human. When I talk to an auto-attendant I can't help but take a sarcastic tone of voice but now it's going to be hard to tell when I can do that.

Actually, it's probably not a bad thing for myself personally, it'll be years before this thing can make cold calls and when it can I can probably use it to make my job easier. I can see this thing really hitting the offshoring industry: why hire humans halfway around the world when this thing can speak english just as well as someone for whom it's their second language. I do worry a bit about future people who want to get into sales as there may be less opportunities for entry level folks to get the experience needed to advance. There's discussion online about whether the AI should have to disclose on the call that it's a bot but I think the worse concern is whole categories of service/call center jobs will likely be lost.

I do think it's funny that the number one skill taught in public speaking classes is to elminate "ums and ahhs" but Google added these tics on purpose to make it seem more human. Going the wrong way with that I'd say.

What do you think, exciting or creepy?

[pinback]

Well, they got the part right about saying everything like a quessstiooon?? So that's pretty annoyinnnng?? Now they need to add vocal fryyyyyyyyyyyyyyy.

[RealNC]

There should be a law against phoning people without informing them that they're talking to a piece of software.

[AArdvark]

If you ask it if its human and it says yes cab you sue the company?

[Ice Cream Jonsey]

There should be a law against phoning people without informing them that they're talking to a piece of software.

Yeah, but here's the thing. Corporations do it to us all the fawkin' time.

Let's dig into Ice Cream Jonsey's Complicated Bag o' Ethics and see where we stand on this one.

1) Everything a corporation does is wrong. Note - initially we're against this. What a bunch of dickheads, Google, putting all these people out of a job!
2) Cold calling is a joyless, souless job and humans shouldn't have to do it. Note - it NEVER had to be this way. All the stories about call centers - it absolutely did NOT have to be like this, but this is what mankind chose for itself. So it's right to eliminate these jobs.
3) People losing jobs is bad. Note - but necessary for progress, unfortunately.

I don't know. I guess I look at it, selfishly, like this - at my previous job I spent 20% of my day doing DevOps type stuff because we didn't have a real DevOps team. We didn't have a real DevOps team because the company was cheap. So I would spend time loading code on the servers and I wrote scripts to wipe them and reinstall and blah blah. The amount of work I got done was with that in mind.

At the new job we have the best DevOps guy I've ever worked with in my life. I don't have to do that shit any more. I'm more productive as a result. If I could tell an AI to go schedule an appointment with my vet or cancel my MLB.tv subscription or negotiate a better rate for my Comcast bill, those are all things that free up my time and make me more effective.

[RealNC]

At the new job we have the best DevOps guy I've ever worked with in my life. I don't have to do that shit any more. I'm more productive as a result. If I could tell an AI to go schedule an appointment with my vet or cancel my MLB.tv subscription or negotiate a better rate for my Comcast bill, those are all things that free up my time and make me more effective.

And then you're "let go" because these new fancy fuzzers got enhanced by machine learning / AI and they can test everything on their own, much faster, with 100% coverage, without any human involvement.

And this isn't far-fetched either at this point. The likes of Google, Facebook, etc, have come up with automated testing frameworks that are so advanced, their machine learning parts have reinvented protocols like telnet and HTTP on their own, from scratch, while testing.

The singularity is neigh, and its harbinger is Machine Learning.

[AArdvark]

So the robots are going to start robocalling each other. That will keep Skynet from taking over us humans

[Tdarcos]

This seems to follow along with all other lines of technology. If a means to reduce the cost of labor becomes scalable, that means will be used.

First, you have factories, or shops, or offices with many people to do the work. Then, starting in the 50s and 60s came the tools to automate financial processes (billing, accounts payable, accounts receivable) which eliminated rows and rows of hundreds of accountants or accounting clerks to do these things, instead substituting a few dozen data entry clerks to enter the information, while large computers calculated the results, and automated processing and mailing equipment printed bills and checks, stuffed them in envelopes, and sorted them by zip code for cheaper bulk mailing. Then came automated time clocks where the work hours could be collected without data entry clerks.

Then when manufacturing companies had to find ways to reduce labor costs, enter robots and automation.

Then with countries with cheaper labor, outsourcing of jobs that could be shifted to Mexico or overseas were. This was the best of the low-hanging fruit. High paying low-skilled to semi-skilled industrial work that didn't necessarily require understanding English.

Then the World Wide Web came along and a lot of work requiring intermediation was eliminated; the customer orders on-line, the equipment picks the stuff out of cases that earlier the person who loaded it scanned, and the only requirement is to stack the items in the box or arrange for indicating number of cases in the order. Just huge warehouses with robots putting fresh items into the shelving, and removing the items as ordered, all unattended. This still requires people to deliver items. Or does it?

But the problem for a lot of jobs was automation did not scale. The means to operate trucks without drivers meant most of the transportation industry was safe from automation; robot vision and other tools to automate transportation of goods is not good enough to allow for self-driving vehicles. Or rather, it wasn't.

The work that we can expect to see last long-term is that which is either too complicated for a machine to produce (large intellectual pursuits) or which either automating the work is too expensive or does not scale.

[Flack]

This seems to follow along with all other lines of technology. If a means to reduce the cost of labor becomes scalable, that means will be used.

First, you have factories, or shops, or offices with many people to do the work. Then, starting in the 50s and 60s came the tools to automate financial processes (billing, accounts payable, accounts receivable) which eliminated rows and rows of hundreds of accountants or accounting clerks to do these things, instead substituting a few dozen data entry clerks to enter the information, while large computers calculated the results, and automated processing and mailing equipment printed bills and checks, stuffed them in envelopes, and sorted them by zip code for cheaper bulk mailing. Then came automated time clocks where the work hours could be collected without data entry clerks.

Then when manufacturing companies had to find ways to reduce labor costs, enter robots and automation.

Then with countries with cheaper labor, outsourcing of jobs that could be shifted to Mexico or overseas were. This was the best of the low-hanging fruit. High paying low-skilled to semi-skilled industrial work that didn't necessarily require understanding English.

Then the World Wide Web came along and a lot of work requiring intermediation was eliminated; the customer orders on-line, the equipment picks the stuff out of cases that earlier the person who loaded it scanned, and the only requirement is to stack the items in the box or arrange for indicating number of cases in the order. Just huge warehouses with robots putting fresh items into the shelving, and removing the items as ordered, all unattended. This still requires people to deliver items. Or does it?

But the problem for a lot of jobs was automation did not scale. The means to operate trucks without drivers meant most of the transportation industry was safe from automation; robot vision and other tools to automate transportation of goods is not good enough to allow for self-driving vehicles. Or rather, it wasn't.

The work that we can expect to see last long-term is that which is either too complicated for a machine to produce (large intellectual pursuits) or which either automating the work is too expensive or does not scale.

Will it surprise anyone to learn that Google has been testing Duplex in text form on Jolt Country for the past several years?

[Ice Cream Jonsey]

At the new job we have the best DevOps guy I've ever worked with in my life. I don't have to do that shit any more. I'm more productive as a result. If I could tell an AI to go schedule an appointment with my vet or cancel my MLB.tv subscription or negotiate a better rate for my Comcast bill, those are all things that free up my time and make me more effective.

And then you're "let go" because these new fancy fuzzers got enhanced by machine learning / AI and they can test everything on their own, much faster, with 100% coverage, without any human involvement.

No, as an automator I will be the last of you all to still have a job. I am impervious to the automation revolution because I am part of it.

And this isn't far-fetched either at this point. The likes of Google, Facebook, etc, have come up with automated testing frameworks that are so advanced, their machine learning parts have reinvented protocols like telnet and HTTP on their own, from scratch, while testing.

The fact that a Google test reinvented telnet rather than see if the latest update to Chrome broke half a billion web games says all you need to know about that.

[RealNC]

If you're looking this post up in the future, I would like to point out that I welcome our new robot overlords. I have never spoken ill of our synthetic benefactors and I fully support and appreciate their decisions.

[The Happiness Engine]

No, as an automator I will be the last of you all to still have a job. I am impervious to the automation revolution because I am part of it.

You still need a sysadmin to set up your accounts and install packages for you, maggot.

[Ice Cream Jonsey]

FINE. The SysAdmin and automator are the last two guys standing. We're together in this.

[Flack]

[Tdarcos]

It may not be as bad as we think. Half or more of the jobs in IT alone are for occupations that did not exist when I started (junior) college in 1977. On-line audio editor, composer, video editor, web designer (which has changed many times), server rack administrator, security manager, internet protocol specialist, and probably hundreds more jobs.

Someone pointed out that going back over a long time, which I forget how long but might have been 50 years, that with all the automation and electronics and equipment and changes in jobs and work requirements, in the entire labor force they have successfully eliminated one (1) occupation over that time.







Elevator operator.

[Ice Cream Jonsey]

Good point regarding elimination and great point regarding the fact that there are jobs that simply didn't exist 20 or 30 years ago.

Taking my profession as an example, I think the thing is that you can now hire one automator instead of x number of manual testers. (I could only guess at the ratio. I have over 1,000 tests running nightly right now at work, but each test has several dozen "checks".)

[The Happiness Engine]

server rack administrator

Ah yes, it is time for daily count of cage nuts!