Jump to content

Welcome to FutureTimeline.forum
Register now to gain access to all of our features. Once registered and logged in, you will be able to create topics, post replies to existing threads, give reputation to your fellow members, get your own private messenger, post status updates, manage your profile and so much more. If you already have an account, login here - otherwise create an account for free today!

These ads will disappear if you register on the forum

Photo

When will androids become physically indistinguishable from real humans?


  • Please log in to reply
14 replies to this topic

Poll: When will androids become physically indistinguishable from real humans? (22 member(s) have cast votes)

Choose a decade for this prediction.

  1. 2020s (2 votes [9.09%] - View)

    Percentage of vote: 9.09%

  2. 2030s (5 votes [22.73%] - View)

    Percentage of vote: 22.73%

  3. 2040s (5 votes [22.73%] - View)

    Percentage of vote: 22.73%

  4. 2050s (2 votes [9.09%] - View)

    Percentage of vote: 9.09%

  5. 2060s (4 votes [18.18%] - View)

    Percentage of vote: 18.18%

  6. 2070-2099 (4 votes [18.18%] - View)

    Percentage of vote: 18.18%

  7. 2100 or later (0 votes [0.00%])

    Percentage of vote: 0.00%

  8. Probably never. (0 votes [0.00%])

    Percentage of vote: 0.00%

Vote Guests cannot vote

#1
wjfox

wjfox

    Administrator

  • Administrators
  • PipPipPipPipPipPipPipPipPipPip
  • 9,199 posts
  • LocationLondon

When will robots cross the 'uncanny valley' and become essentially the same as humans in appearance?

 

https://en.wikipedia.../Uncanny_valley

 

 

JOTEkTr.jpg

 

 

 

I realise we already have some pretty realistic-looking androids – but you can still tell fairly easily that they're machines, especially when they move and speak. The kind of robots I'm talking about would be essentially the same as "Data" from Star Trek: TNG, or the female robots in Ex Machina, or Kara from Quantic Dream.

 

 

 

 

 

 

I'm in the process of re-writing a prediction, plus adding a couple of news ones...

 

http://www.futuretim...status_id=20887

 

... so I need your opinions on this. Please vote in the poll above. :)

 

Thanks!

Will


  • Yuli Ban likes this

#2
Raklian

Raklian

    An Immortal In The Making

  • Moderators
  • PipPipPipPipPipPipPipPipPipPip
  • 6,782 posts
  • LocationRaleigh, NC

If we successfully print equivalents of skin and hair (as well as exterior body parts), it won't be too long before we coat a robot with an appearance that is identical to that of a human being.

 

I think in a decade or two, Boston Dynamics and its subsequent spinoffs will be able to have robots animate with human-like movements. 

 

To top it off, a team of Columbia University researchers have designed a 3D printed silicone rubber matrix that is capable of mimicking human muscle that is actually three times stronger than the human counterpart.

 

There is no reason we won't be able to achieve a life-like android/gynoid in terms of outward appearance within 20 years.

 

However, for that android to develop a form of consciousness that will make it think like a human.... that's a whole different story. It make take another century or so.


What are you without the sum of your parts?

#3
Yuli Ban

Yuli Ban

    Born Again Singularitarian

  • Moderators
  • PipPipPipPipPipPipPipPipPipPipPip
  • 19,150 posts
  • LocationNew Orleans, LA

Some aspects of total realism are already possible. One thing that's almost never mentioned about these androids and gynoids is that they're working on university budgets. You can either work on a practical robot that can do things or you can work on a robot that looks realistic, and most go for the former. Some of these androids might have been created with budgets as low as $20,000 to $50,000. 

 

Now I want you to compare that to Hollywood animatronics, which are given millions of dollars for construction and are constantly refurbished. 

 

 

It's not perfectly out of the valley— the eyes are kinda off and the way the jaw moves just has a bit of uncanniness to it, but when I orient it right and obscured the neck, my brain tolerated it much more than this:

It's not out of the valley, but it's far up the slope. Sophia, on the other hand, triggers my fight-or-flight response.

 

One of the reasons why this one's so fantastic is because of the gap between entertainment and science funding.

 

Thing is, prices come down over time once you need fewer hands to build this thing, and I think it was Starspawn0 who said that it would be AI itself that will bridge the gap between the Uncanny Valley and Total Realism. 

 

One of the things about natural intelligence that separates it from artificial intelligences is just how fucking generalized it is. It's so stupidly generalized that we don't even realize it. We look at things like being able to open a door and drive a car as examples of general intelligence, but our brains are also unconsciously controlling our internal organs, muscles, nerves, et al while our consciousness persists. This masterful control of muscles is what stops humans from looking uncanny. It's one reason why those with extreme mental deficiencies sometimes cross into the Uncanny Valley— their motor and muscle control is greatly reduced, and it looks unreal to us. Whereas robots barely have motor and muscle control at all. Once you have the brains, the rest of it comes naturally. You just need the body then, and thus you need the right muscles or substitutes. Without a strong-enough brain, even a perfect muscular recreation of a human face will act uncannily. 

 

 

When will this occur? Well it really depends on when we get robots that can control a body and all its muscles. I don't doubt that we can do it and possibly even soon— computers today are stupendously strong, just inefficient compared to the human brain. It's not like basic domestic utility robots where we already have absolutely everything we need but lack the brain power. The muscles to make totally realistic faces on a small scale are still lacking as well, so there's more to be done in that field. But for the most part, it's an AI issue.

 

I'm going to say we'll see at least a head that looks and acts physically indistinguishable from a real human by the 2030s. Full body might take until the 2040s. But who knows, maybe it'll be the opposite, because we are evolutionarily optimized to notice facial motion more than anything after all. So 2030s-2040s is about right.

 

 

 

 

Now when am I gonna be able to lay down that gynoid waifu? Hmm, I can't say, because I'm thinking about this in the same manner as the way we view great achievements like those photorealistic renders from a few years ago. Things that we managed to do with enough time, effort, and power, but certainly not something you can do with a $500 computer. It'll likely take a bit longer for realistic androids to diffuse throughout society in any real number because I can't see them being anything less than astronomically expensive for a time.


  • eacao and Casey like this

And remember my friend, future events such as these will affect you in the future.


#4
Yuli Ban

Yuli Ban

    Born Again Singularitarian

  • Moderators
  • PipPipPipPipPipPipPipPipPipPipPip
  • 19,150 posts
  • LocationNew Orleans, LA

Another thing I wanted to mention in its own post: it seems as bizarre to say androids will be indistinguishable from humans in 30 years as it must have been to say "we'll put a man on the moon within the century" in the 1910s. Indeed, just as airplanes of the 1910s were sad little things that often flew slower than most modern cars can drive and rockets hadn't fundamentally changed since the days of the Song Dynasty, modern robots are still fundamentally similar to what we were creating 50 years ago, but with much more brainpower, moving parts, and efficient designs. 

 

One of the ways you can tell just how primitive robotics are actually comes from when robots fail. Have you ever seen a robot fall over? Like the famous video of ASIMO pratfalling on stairs?

 

Nowadays, we have backflipping robots. But they still fall like robots. They completely freeze and plop. Real people (usually) don't do that. We fall "gracefully"— we try to catch our descent, or our limbs flail, and our body parts bounce with some level independence from each other (leg might react in a different way from arm; fingers tense up and the opposing hand's fingers are outstretched to reach something). Robots fall robustly— every part of their body reacts the same way to the fall and impact. 

 

It's something people don't think about but creating robots that fall down gracefully will actually go a very long way towards getting androids out of the Uncanny Valley.


  • wjfox likes this

And remember my friend, future events such as these will affect you in the future.


#5
dekoomer

dekoomer

    Member

  • Members
  • PipPip
  • 15 posts

Even if androids looks just like a real person, The uncanny valley can still be there, If their behavior is a bit off or not quite to the level of an average human being. Or if you're having androids doing tasks at a super human speed (moving, speaking, ect).

 

IMO beside the porn industry. Society may prefer having androids that very much easy to spot and recognize as being androids, It may take the dawn of the next century before the masses of the world are comfortable with androids that looks like us.

 

However in places like japan and south korea they will be ahead of the world in the acceptance of human like androids (This is likely where our predictions would come true). 


  • wjfox likes this

#6
kjaggard

kjaggard

    Artificer

  • Moderators
  • PipPipPipPipPipPipPipPip
  • 2,891 posts
  • Locationwhere fanciful imaginings and hard won knowledge meet to genesis the future.

that's a tricky one. there are some artists that create hyper realistic statues of human beings that are mind boggling, but that's stationary, when you manage to get the motion and physiological changes while in motion correct you are still going to has a sense of something off about one because there are micro shifts in under skin structure that make the human body always in motion even when sitting still the pupils in the eye, the pulse through arteries, the slight movement of nostrils and lips in breathing, human cheeks move with changes in pressure when a person swallows their saliva. And that's not even counting the hundreds of varieties of micro ticks and twitches that humans perform as just part of a body in motion while a mind performs background tasks, these are aspects of a persons personality that we rarely notice (think of tells in a game of poker) but we feel something is off when they are absent.

 

If you make one that is meant to be an unnoticed person, that nobody looks at or interacts with but is there, doesn't speak (something like a door man, limo driver waiting by the car or secretary in a meeting), It would pass the test.

 

But I'm more of the opinion that aiming for looking exactly like a human is the wrongs direction. We have fantasy with orcs, dwarves, satyrs and mermaids. we can accept the idea of people that are not quite human, and we can even accept that they have abilities that they are better at and designed for that we aren't. Why do we have this conceit that the human form is the direction in which robots must go?

 

Even when going humanoid to tap into the environment and abilities that afford them in a word built to the human form, why four fingers and a thumb instead of three fingers and a thumb, or 9 small prehensile tendrils on each hand? Why two arms and not four? Why not digitgrade legs? An array of six eyes?

 

Alright fine, two arms, two legs, humanoid torso, head on neck, two eyes etc. The iron man armor matches that description, so does c 3po.

 

I guess what I'm saying is, given a new type of being, why make it conform to the patterns of something that already is there. We can already create more beings indistinguishable from humans, they are called offspring. 


  • Jakob likes this
Live content within small means. Seek elegance rather than luxury, Grace over fashion and wealth over riches.
Listen to clouds and mountains, children and sages. Act bravely, think boldly.
Await occasions, never make haste. Find wonder and awe, by experiencing the everyday.

#7
Jakob

Jakob

    Groveling Goblin

  • Members
  • PipPipPipPipPipPipPipPipPipPip
  • 5,931 posts

Well for the main applications I imagine for androids--personal assistants/companions and child and elderly care--it would probably be best if they were as close to human as possible. You wouldn't want a bunch of kindergartners to be looked after by a six-eyed man with tendrils and metallic skin, they'd get scared.


  • wjfox likes this

#8
wjfox

wjfox

    Administrator

  • Administrators
  • PipPipPipPipPipPipPipPipPipPip
  • 9,199 posts
  • LocationLondon

The poll results are showing a very mixed opinion – was hoping for some kind of consensus on this topic. :) This is going to make it harder for me to write about. Perhaps if I go for the middle range (2050-ish), then I can incorporate both a pessimistic and optimistic viewpoint, somehow.



#9
Alislaws

Alislaws

    Democratic Socialist Materialist

  • Members
  • PipPipPipPipPipPipPip
  • 1,354 posts
  • LocationLondon

I went with 2030s, As that (amazing) animatronic head shows, we can make realistic looking people now, what we need is for the price and effectiveness of artificial muscles to come down.

 

Once we have a physical robot which is capable of the full range of human facial motion, (only needs 42 muscles!) using machine learning techniques to train an AI to move the face realistically will be easy, it will have millions of hours of footage human faces moving realistically to compare it's work with, if that isn't enough we can get volunteers to wear sensors to track muscle movements in real human faces and give it that data. 

 

I have no idea what the global market is for prostitution/pornography/sex trafficking/slavery...* is but it is probably a lot of money, and producing a realistic Gynoid/Android would allow you to grab a big chunk of that money. 

 

*Anything that involves people satisfying themselves by having other people do horrible, or unpleasant, tasks would fit really, I'm sure there will be people using them for "deathmatch" boxing/MMA (controlled remotely by human athletes using VR maybe?) and Westworld style theme parks when the AI gets more advanced. Humans have some serious primitive drives around social hierarchy, and being above other people etc. so demand for "people" to lord it over in a morally neutral way will be high. 



#10
funkervogt

funkervogt

    Member

  • Members
  • PipPipPipPipPip
  • 414 posts

I think making an android that has a hyper-realistic human-looking body will be the easy part, and could probably be done today. However, the much harder part is designing the robot to accurately mimic human physical motions, particularly fine motor movements, walking smoothly, and overcoming the uncanny valley in its facial expressions. I don't expect all of it to come together until near the end of this century. 



#11
Alislaws

Alislaws

    Democratic Socialist Materialist

  • Members
  • PipPipPipPipPipPipPip
  • 1,354 posts
  • LocationLondon

Its possible that building musculo-skeletal systems that can mimic human performance will be hard but i'd be very surprised if AI can't animate a human face by 2030, we can keep progressing with that even without the robots, since its basically the same as creating digital animations. We can already take human face input and produce a pretty good animation set from that. 

 

It may be a long time before an AI can mimic a human well, but we should be able to map human performances to an android and produce something realistic by the late 2020s/early 2030s I think. 



#12
funkervogt

funkervogt

    Member

  • Members
  • PipPipPipPipPip
  • 414 posts

Its possible that building musculo-skeletal systems that can mimic human performance will be hard but i'd be very surprised if AI can't animate a human face by 2030, we can keep progressing with that even without the robots, since its basically the same as creating digital animations. We can already take human face input and produce a pretty good animation set from that. 

 

It may be a long time before an AI can mimic a human well, but we should be able to map human performances to an android and produce something realistic by the late 2020s/early 2030s I think. 

Humans are exquisitely attuned to reading the faces of other humans, and I doubt we'll build an android face that can correctly and accurately reproduce all the subtleties of human expression until well after 2030. 


  • wjfox and Casey like this

#13
Alislaws

Alislaws

    Democratic Socialist Materialist

  • Members
  • PipPipPipPipPipPipPip
  • 1,354 posts
  • LocationLondon

We just need to mimic the functionality of human muscles, then we have the human face as a model for how to arrange the 42 muscles on a skull which can just be an exact replica of a human's. I've seen a lot of headlines about breakthroughs in artificial muscles over the last couple of years, but it remains to be seen how close those can get to biological muscle performance in their commercialised form.

 

As soon as we have the muscles, then we can make androids which can move how we do. The software side of the problem will be solved by people in gaming and movie special effects etc. long before the hardware is ready to produce a human looking android but I don't think artificial muscles are a technology we will have to wait till 2040 for. I could be being too optimistic on that though. 

 

EDIT: Articles like the following are why I think we will have this solved before 2040. (also contains a video of the muscle in action) They don't address how fast the muscle can contract though, so I suspect this product won't be as useful as the article says. 
 

 

https://www.engadget...t-breakthrough/

 

Synthetic muscle breakthrough could lead to 'lifelike' robots
 
Researchers claim it's the closest artificial material equivalent to a natural muscle.
...
Researchers at Columbia Engineering have developed a 3D printed synthetic tissue that can act as active muscle. The material, which can push, pull, bend, and twist (thanks to its use of silicone rubber and ethanol-dispensing micro-bubbles) is also capable of carrying 1,000 times its own weight.
...
"This is a big piece of the puzzle and, like biology, the new actuator can be shaped and reshaped a thousand ways. We've overcome one of the final barriers to making lifelike robots."...

 



#14
Casey

Casey

    Member

  • Members
  • PipPipPipPipPipPip
  • 613 posts

I think making an android that has a hyper-realistic human-looking body will be the easy part, and could probably be done today. However, the much harder part is designing the robot to accurately mimic human physical motions, particularly fine motor movements, walking smoothly, and overcoming the uncanny valley in its facial expressions. I don't expect all of it to come together until near the end of this century. 

 

I dunno, I don't think the problems you listed sound difficult enough to take that long to overcome. My guess would be sometime during the 30s or 40s. I voted 2040s.



#15
funkervogt

funkervogt

    Member

  • Members
  • PipPipPipPipPip
  • 414 posts

http://www.marcsijan.com/gallery/

 

 

I think making an android that has a hyper-realistic human-looking body will be the easy part, and could probably be done today. However, the much harder part is designing the robot to accurately mimic human physical motions, particularly fine motor movements, walking smoothly, and overcoming the uncanny valley in its facial expressions. I don't expect all of it to come together until near the end of this century. 

 

I dunno, I don't think the problems you listed sound difficult enough to take that long to overcome. My guess would be sometime during the 30s or 40s. I voted 2040s.

 

Well, I disagree. 

 

The hyperrealistic sculptures of Marc Sijan convince me we could build android bodies that look externally identical to humans now. However, endowing them with the ability to fluidly move around like humans and to be masters of human facial expression will be a much harder nut to crack. I voted for 2070-2099, but I admit that as a general rule, I err on the side of being conservative in my predictions. 

 

http://www.marcsijan.com/gallery/






0 user(s) are reading this topic

0 members, 0 guests, 0 anonymous users