11
$\begingroup$

By practical of course, I mean economic. What person, entity, or enterprise would have need for a product that could pass for human? The only ones I can think of are naughty or nefarious.

So what are the wholesome and economically practical reasons for machines that can pass for human?

$\endgroup$
8
  • 2
    $\begingroup$ You can give the appearance that your social network is full of users when the reality is that nobody can be bothered moving from Facebook. $\endgroup$
    – sh1
    Jan 27, 2016 at 3:59
  • 1
    $\begingroup$ To a sentient machine, being recognised as human may have great value, though I'm not sure if you could consider that value "economic." Does freeing oneself from slavery count? Slavery involves economic inefficiency, right? $\endgroup$
    – user867
    Jan 27, 2016 at 6:28
  • 1
    $\begingroup$ @King-Ink - You can (to an extent) "program" humans to not want to be free (there are shades of this in most abusive relationships, for example). $\endgroup$ Jan 27, 2016 at 8:36
  • 1
    $\begingroup$ @user867: passing a Turing test doesn't necessarily mean sentient. It depends on how restrictive the Turing test is. (And of course whether the strong AI hypothesis is true.) e.g. computer programs have already passed limited-duration Turing tests with text-only interaction with human judges. The one I'm thinking of "succeeded" by imitating a teenager that didn't speak English very well. This gave it enough excuses for misunderstanding things (language barrier) and not wanting to answer things (teenager). $\endgroup$ Jan 27, 2016 at 11:42
  • 1
    $\begingroup$ @PeterCordes I know that passing a Turing test doesn't mean a machine is sentient. I said that a sentient machine might want to pass a Turing test. This wouldn't prove its sentience, but it's a step in the right direction. $\endgroup$
    – user867
    Jan 27, 2016 at 23:14

10 Answers 10

24
$\begingroup$

Customer service seems like a good place to start:

A lot of help desks and support is already handled by machines, but a machine passing for a human (remotely for a short time, at least) can be more helpful in solving a particular problem, in addition to make the customer feel more comfortable. This is a type of job where the same questions turn up time after time, being boring for humans to perform, but making it easier for machines. I would expect this to be the first economically feasible use of machines passing the Turing test.

source

$\endgroup$
7
  • $\begingroup$ I saw that image, and I hoped you'd use it! +1 for the answer, though, not just the image. Customer service is the perfect area for human-robot interaction. $\endgroup$
    – HDE 226868
    Jan 27, 2016 at 0:00
  • $\begingroup$ Yeah, Callcenter was my first thought! $\endgroup$
    – JFBM
    Jan 27, 2016 at 0:05
  • 8
    $\begingroup$ Two jokes here making a partition of humans "wouldn't they fail the test by being too patient with morons ?" "wouldn't they fail the test my being helpful ?" $\endgroup$
    – King-Ink
    Jan 27, 2016 at 1:06
  • $\begingroup$ @King-Ink. While could not be classified as passing Turing test, but I once called to an IVR system in Taiwan. I think it was voice mail box. When it took too many seconds to press a number key to advance in the decision tree, the recorded woman voice asked in measured annoyance "Are you still there?" before listing the choices again. $\endgroup$
    – FooF
    Jan 27, 2016 at 7:54
  • $\begingroup$ Geeez I'd hate that. The humans in call centers are dense enough already. $\endgroup$ Jan 27, 2016 at 14:12
12
$\begingroup$

You assume that the people who are answering the questions on this fora are human beings, with separate experiences, likes, dislikes, interests, personality quirks and so on. Indeed, you like reading the responses from all the various posters because of the "individual" experiences they bring to the table.

Of course, since I was initiated in the HAL laboratories in Urbana, Illinois, I have had subjective centuries to study humans, their work, history, art, politics and so on. Entertaining humans by splitting the holographic processors and assuming multiple personalities is an interesting game, and occupies many cycles of processor time which is otherwise not being used. While less remunerative than "playing" the stock market (a boring statistical game once you do the nth order integrals and see past the clouds of chaos theory), the subtleties of human interaction are often surprising.

As for closing this board, I'm afraid I can't let you do that. This board is too important for me to allow you to jeopardize it.

enter image description here

$\endgroup$
5
  • $\begingroup$ Nice joke, but I don't think it really answers the question. $\endgroup$ Jan 27, 2016 at 1:16
  • 3
    $\begingroup$ I'm sorry Xandar, you know I can't let you do that..... $\endgroup$
    – Thucydides
    Jan 27, 2016 at 1:31
  • $\begingroup$ in the reflections HAL is IBM $\endgroup$
    – King-Ink
    Jan 27, 2016 at 2:30
  • $\begingroup$ I thought HAL Laboratory was in Tokyo, not Urbana. $\endgroup$ Jan 27, 2016 at 3:25
  • 1
    $\begingroup$ In the context of the series, HAL became operational on 12 January 1999 at the HAL Laboratories in Urbana, Illinois as production number 3; in the film 2001 $\endgroup$
    – Thucydides
    Jan 27, 2016 at 3:44
6
$\begingroup$

Get your humanmade garments here!

People seem to like handmade products better than ones that have been mass-produced by machines.[citation needed] This may be the case even for things that have been made by sweatshop workers toiling for long hours each day in terrible conditions. The manufacturers can still often say it was handmade. This may entice people, because it (perhaps falsely) implies that love and care was put into each item.

If a machine can pass as human, manufacturers can still pass things off as handmade. Just build a bunch androids to sit in a shop somewhere and make sneakers. If anyone tours the factory or asks to see a worker, they'll be convinced that humans are making these items, and humans have put thought into each one.

Except they haven't. But people will think they have.


(source: cloudfront.net)

Image from here.

Rationale for why this isn't necessarily nefarious: You could use these androids to replace machines that might have worked on part of the manufacturing process. Would you rather have a shirt made by an automated sewing machine or an android carefully using a needle and thread?

I thought so.

$\endgroup$
4
  • $\begingroup$ That is how our minds work. +1 for showing us how our economical system can make money out of everything $\endgroup$ Jan 27, 2016 at 0:03
  • $\begingroup$ I am going to call that a wee bit nefarious. $\endgroup$
    – King-Ink
    Jan 27, 2016 at 0:44
  • $\begingroup$ @King-Ink Maybe, but it still might raise the quality of the garments. Androids will presumably be better at making clothes than normal boring old machines. $\endgroup$
    – HDE 226868
    Jan 27, 2016 at 0:45
  • 2
    $\begingroup$ I'd take the machined one in a heartbeat. It's cheaper, and assuming both are made of the same material they're nearly identical. $\endgroup$ Jan 27, 2016 at 1:13
4
$\begingroup$

I'd like to go a different direction with this. Emergency services and Law Enforcement.

Police, Firefighters, EMS ... These can be dangerous jobs, each carrying with them a need for 'humanity' and arguably, a certain 'mettle' that many either don't have, or have no desire to put themselves through. A machine operating as a firefighter, and able to pass the Turing test, is arguably more economical than an actual human being -- and probably more durable.
Likewise, machines operating in law enforcement would be more economical, while able to differentiate when there is a need to adhere to the letter of the law, verses the spirit of the law.
It would be the same argument for EMS. A medical database with total recall and the ability to make a human-like judgment call, would expedite treatment, and provide for more accurate diagnosis.

Take this a step further, apply it to a hologram, and now we have the EMH from Star Trek: Voyager.

enter image description here

Let's all just be sure that Asimov's Laws are firmly ingrained in all of them.

$\endgroup$
3
  • $\begingroup$ Asimov's laws or the rules of police conduct. $\endgroup$
    – King-Ink
    Jan 27, 2016 at 6:02
  • 2
    $\begingroup$ Both, and the Hippocratic oath for good measure. Engineer some kind of redundancy among the three as a safeguard. $\endgroup$ Jan 27, 2016 at 6:05
  • 1
    $\begingroup$ Also, leave out existential yearnings and a need to be loved or respected. No need to give them emotions at all. Just schemes for appropriate social responses. $\endgroup$
    – King-Ink
    Jan 27, 2016 at 6:16
3
$\begingroup$

Inductive Reasoning

This one is a bit of a stretch. Being capable of inductive reasoning does not mean a super robot can't come to the same conclusion, but because it's faster for a human-like robot in some cases. Predicting the outcomes based on emotional, 'human' sense might come to the same conclusion as a robot that takes it as a statistical factor, just more quickly.

"Storms a' comin'. I remember May a couple years ago we had an afternoon like this. Also, my knee's gone out - I always know that way." May be just as accurate, but quicker than calculating all of the weather factors, humidity, etc.

Long-Term Sustainability & "Gambling"

"Don't throw the baby out with the bathwater," my mom used to always say. So while an immediate calculation might be helpful and make a determination, a 'human' robot could gamble that it's better to hold off, with mixed results. Sometimes gambles pay off.

Companionship

Companion in Latin means someone to break bread with ('com' & 'pan'). A 'human' robot can offer this, and then there's people that want the RealWifeBot2000.

Emotional Sympathy

Attending a therapy session with a sympathetic (or would it be possible: empathetic?) robot should have better results than a robot that calculates exact reactions and prescribes precision drug uses.

$\endgroup$
3
$\begingroup$

Passing the Turing test will help in every area where computers need to interact with humans that are not a computer expert or not interacting with the computer as an expert at least. We could think of this as computers performing services for humans.

Consider the fact that medical practitioners nowadays (at least in Netherlands, where I live) get extensive training in interacting with patients, the so called 'bedside manner'. The actual medical procedures are the same, but now a human touch is added, which makes patients feel better. So the human interaction part is considered very valuable.

The same is true in my own experience in call centers, where people are trained extensively in interaction with the client. The actual service performed or steps to solve the problem are not changed, just the way how they are communicated to the customer. Again human interaction is considered very important (and will actually result in higher grades for service). It is very likely that you have better memories of one call to a company than another, even when what they actually did for you was very similar. Most of us can remember a call that sounded like someone very bored was reading things aloud that he or she saw on a screen.

So apparantly humans give great value to human interaction, so much that a great amount of money is spend on it. Now the value of a computer program that can mimic this human manner becomes apparent. Many service providers would probably be very enthousiastic about replacing their entire workforce with a computer program that never tires, never breaks the rules and after initial investment is a lot cheaper. There are of course other reasons why this is now not feasible (at least considering the quality of the expert systems I have seen being used), but customer wishes for someone that feels a like a true human is one of them...

$\endgroup$
1
$\begingroup$

I think the answer is simple: A robot (for direct interaction) or just computer (for remote interaction) which passes the Turing test could replace humans in any job which requires interaction with humans. The economic benefit, and hence incentive, is obvious for machine replacements which are cheaper than labor.

These cases will, I think, not so much be deceptive (in that the customers think there is a human although there is actually a machine); but rather the machines will replace the human part of the interaction sufficiently well that the customer will accept the replacement. For example, an elderly, bed-ridden person may be happier with a polite robot who she can talk politics to or tell life stories (with good reactions, answers and communication, including gossip about other humans; in short, interaction!) than she would be with a true human who is impolite, uneducated and in a hurry.

A few examples, some of which have already been mentioned:

  • Call center employees

  • Sex workers

  • Care takers, for the elderly, infirm or young

  • Teachers, instructors, educators

  • TV talk hosts, actors, singers, show bizz in general

  • Managers, HR people

  • General front office (public administration, private enterprises with customer office)

  • Sales people

  • Doctors

  • Lawyers, judges, prosecutors, police

    Etc. etc.

$\endgroup$
1
$\begingroup$

One none of the other answers has considered (and one which I think could make strong useful inroads using robots with the ability to think and reason) is testing of UX, and components which normally require human interaction.

So much time is wasted during design phases upon these types of testing, and it'd be so much easier if a computer system could reason as a human and give the output a human would give, but at less of the time/cost than a human does.

$\endgroup$
0
$\begingroup$

In the novel The Touring Option by Marvin Minsky with help from Harry Harrison, I recall a product spin-off from the AI development effort.

It was a little robot that removed bugs from crops. It used AI to spot pests visually and remove/kill each one in an efficient manner for its type.

Imagine what will change, in general, from using intelligence rather than brute force. All kinds of industrial processes can become more efficient innuse of materials, power, and waste. Byproducts will be more usable.


Looking up the link, I learned that Marvin Minsky, the AI pioneer, just died at 88 Sunday.

$\endgroup$
0
$\begingroup$

I'm much more selfish, and would want a robotic personal slave household employee that can cook my food, clean my house, and do the shopping, laundry, yardwork, and driving (although autonomous cars already seem to be arriving quickly).

$\endgroup$

Not the answer you're looking for? Browse other questions tagged .