The skeptic’s guide to humanoid robots going viral on the Internet
- Thread starter JournalBot
- Start date
That kid in the white t-shirt and blue shorts on the left looks like one of the people in other such viceos that have/use the robot's contoller behind their back...
ETA: Forgot to mention I saw it happening on a serpentza and c.milk video about chinese robot scams
ETA: Forgot to mention I saw it happening on a serpentza and c.milk video about chinese robot scams
Upvote
-3
(6
/
-9)
Seems there are two things being conflated:
Although, I could see a use for humanoid robots that could be teleoperated. Most consumer-oriented aerial drones aren't autonomous and yet are still useful. A mechanically sound teleoperated humanoid drone could be useful for hazardous environments for example.
- What the robot can do mechanically. Does it have the degrees of freedom, power, endurance, precision, etc to perform in human-oriented spaces?
- What the robot can do autonomously. Regardless of the hardware, if the software isn't up to handling the job and environment safely and reliably the use of the robot is limited.
Although, I could see a use for humanoid robots that could be teleoperated. Most consumer-oriented aerial drones aren't autonomous and yet are still useful. A mechanically sound teleoperated humanoid drone could be useful for hazardous environments for example.
Upvote
93
(94
/
-1)
The anthropomorphism is strong. People see a humanoid robot doing a demonstration (eg, dance moves) and think "neat!" but then get that robot, and tell it to do the same dance moves on ice, or on a slope, or facing the other direction.... and it might not perform so well. Mechanically, it might or might not be capable, but the autonomous control is going to be different as well, and might not be optimized for the environment.
Upvote
8
(16
/
-8)
Very true! Saw another video in one of comments on Ars (the live video of the sorting robot?) that had an interesting video of Boston Dynamics' robots. Unfortunately, it looked like each time they moved the small firdge, they had to reprogram the whole path and distances. No autonomous workflow, although it did appear to meet all the former requirements.
Would it meet the latter's requirements if it was remotely controlled via AI? Would be an iteresting test, especially if multiple robots were being controlled simultaneously.
Edited for clarification
Upvote
4
(5
/
-1)
It depends on the servos. Most are not "heavy duty", but are calibrated and rated for the specific task that they'd be doing. Since, other than standing, those tasks generally are the same for humans as for them, probably not crushing.
And if they're designed for working with PEOPLE, then they'd have to have pressure feedback sensors in their hands to prevent damaging humans.
Granted, you COULD make one into a human killing machine, but if you do that, you'd end up having crowds of short tempered people with large construction vehicles and no sense of empathy for the robots introducing them to an alternate shape of reality.
Upvote
5
(7
/
-2)
If I Imade robots to kill people, nearby direct action is not how I would do it .
Upvote
0
(7
/
-7)
Yes. We clearly understand the purpose of the PR stunts is to fool people who don't know better. Did you at least skim the article?
Upvote
71
(75
/
-4)
Upvote
87
(87
/
0)
He wants a company to build a second robot for the Chicago market that would slap those hot dogs away from the ketchup-dispensing robots.
Begun, the Ketchup Wars have.
Upvote
40
(40
/
0)
Resistance
Wise, Aged Ars Veteran
Look at any automation problem we have solved at any point throughout history and ask yourself:
This suggests that the applications of humanoid robots will be quite limited.
You would also do well to look at what things can be done ridiculously cheap and efficient using automated tools that we still use humans for on a large scale. Human labor, on average, is so absurdly cheap that only labor that must be done in areas with strong labor laws will be replaced by humanoid robots any time soon.
You can also expect that these robots will be operated and trained by what are functionally slaves, you may as well just put a robot costume on a poor person and call it a robot, same thing with less steps.
The answer is, almost universally, no.
This suggests that the applications of humanoid robots will be quite limited.
You would also do well to look at what things can be done ridiculously cheap and efficient using automated tools that we still use humans for on a large scale. Human labor, on average, is so absurdly cheap that only labor that must be done in areas with strong labor laws will be replaced by humanoid robots any time soon.
You can also expect that these robots will be operated and trained by what are functionally slaves, you may as well just put a robot costume on a poor person and call it a robot, same thing with less steps.
Upvote
25
(27
/
-2)
Resistance
Wise, Aged Ars Veteran
Any humanoid robot that can be teleoperated and is cheap enough will be used to do menial labor while paying the operator a wage so low they are functionally a slave.
Upvote
6
(8
/
-2)
Frankly, this and adjacent industries are operating at such a deficit of faith that I don't care what they "mention"--I will hold my horses until I see a demonstration organized and operated by a third party.
Upvote
31
(31
/
0)
Sorceror, actually.
If they can manage that I have another task. They must carry a tray of filled cocktail glasses up a narrow staircase in a bar while drunk patrons are coming down.
Upvote
9
(9
/
0)
A few points - there's also outright fraud. I'll just leave this link here, without any comment, or any connection, necessarily, to my previous sentence: https://www.forbes.com/sites/johnbb...set-fail-video-is-blowing-up-on-social-media/
Then there's the classic issue which almost all difficult problems encounter - the first X% of functionality/usability/whatever is "easy", the next Y% is much harder, and takes much more effort/time/resources to "perfect", and so on until you get to the last 5 or 10%, which may be virtually impossible to ever achieve... in other words, it's an asymptotic curve.
Of course, average schmos don't really get the concept, so they see a robot demo - where the robot may be, say, 80% of the way towards "completion" (ie. "good enough" for public release), and think that we'll all have robot friends in the next year or so, when that next 10% towards "good enough" might take double, triple, or even longer to develop than the whole development cycle to date, and then the next 5% same again and so on.
And, yes, advances in tech might shave some time off that, maybe, sorta... but it's deeply misleading if not downright fraudulent to seek investors' money on the basis that 'we'll be finished soon - just look at how far we've already come'.
Going back to Tesla... just think of Musk's FSD bullshit. Every year, he'd come and out claim that FSD is right around the corner, next year your car will drive itself coast to coast, and other such bullshit. Most recently, I read that they're fraudulently altering the contracts of buyers who were (contractually) promised FSD - https://www.msn.com/en-us/news/tech...spark-evidence-tampering-concerns/ar-AA24Lsyg
[It's also a good lesson to always keep your own copies of important documents, and never trust anyone - especially the other party! - to keep it for you.]
Then there's the classic issue which almost all difficult problems encounter - the first X% of functionality/usability/whatever is "easy", the next Y% is much harder, and takes much more effort/time/resources to "perfect", and so on until you get to the last 5 or 10%, which may be virtually impossible to ever achieve... in other words, it's an asymptotic curve.
Of course, average schmos don't really get the concept, so they see a robot demo - where the robot may be, say, 80% of the way towards "completion" (ie. "good enough" for public release), and think that we'll all have robot friends in the next year or so, when that next 10% towards "good enough" might take double, triple, or even longer to develop than the whole development cycle to date, and then the next 5% same again and so on.
And, yes, advances in tech might shave some time off that, maybe, sorta... but it's deeply misleading if not downright fraudulent to seek investors' money on the basis that 'we'll be finished soon - just look at how far we've already come'.
Going back to Tesla... just think of Musk's FSD bullshit. Every year, he'd come and out claim that FSD is right around the corner, next year your car will drive itself coast to coast, and other such bullshit. Most recently, I read that they're fraudulently altering the contracts of buyers who were (contractually) promised FSD - https://www.msn.com/en-us/news/tech...spark-evidence-tampering-concerns/ar-AA24Lsyg
[It's also a good lesson to always keep your own copies of important documents, and never trust anyone - especially the other party! - to keep it for you.]
Upvote
39
(39
/
0)
It's like spray paint: the trick is to start past the end of the work, and finish past the end of the work, with slow, even motion along the axis of application.
Now, if I could get a robot to remind me to shake it and the mustard first so I don't end up with raw liquid ketchup and mustard juice, we'd be on to something.
Upvote
15
(15
/
0)
Let's be honest here, even humans have issues with everything you listed unless they're trained to do it specifically... I know how to do a lot of dexterous things on flat ground, definitely can't do the same shit on a slope, ice, and etc...
Upvote
-10
(2
/
-12)
Just going to leave this here:
View: https://bsky.app/profile/jjvincent.bsky.social/post/3mayddynhas2l
View: https://bsky.app/profile/jjvincent.bsky.social/post/3mayddynhas2l
Upvote
49
(49
/
0)
Given it's probably just replaying a canned sequence of movements, defeating it would be simple... just walk behind it and push it over.
Upvote
16
(16
/
0)
Bastet's pet Morlock
Seniorius Lurkius
A cheap humanoid robot is cheap enough to risk instead of a human. They can fall off cell towers, strain their backs lifting patients, and disregard smoke inhalation much more cheaply than the people currently doing that. They also can't catch ebola/covid/whatever the hell else requires an overpressure hazmat suit, which might have come in handy a few years back when we lost an enormous quantity of skilled medical personnel.
Upvote
-7
(6
/
-13)
What do you mean by "remotely controlled by AI"? An AI that is connected to the robot in real time via the robot's onboard sensors effectively inhabits it as a body. It would be like saying that your body is remotely controlled by a human mind.
I suppose if an AI were controlling the robot by observing it externally from a third person perspective then it could be said to be remotely controlling it.
Upvote
-12
(3
/
-15)
Many people take technology for granted.
They routinely watch television, but haven't the slightest idea how one works.
Many don't know what the EM spectrum is, have never heard of Ohm's Law, have no idea what a centimeter is.
Some don't know what pi is.
They accept the technology all around them, almost like magic.
It's completely unsurprising that numerous people extrapolate viewed humanoid robotic behavior, particularly decision making capability.
They routinely watch television, but haven't the slightest idea how one works.
Many don't know what the EM spectrum is, have never heard of Ohm's Law, have no idea what a centimeter is.
Some don't know what pi is.
They accept the technology all around them, almost like magic.
It's completely unsurprising that numerous people extrapolate viewed humanoid robotic behavior, particularly decision making capability.
Upvote
21
(23
/
-2)
Fair enough, let's use operating a machine, vehicle, or system from a distance. Or maybe remotely operated as in tele-operated with no external physical connections, using the onboard sensor reading and a remote AI server, air-gapped or otherwise separated from the physical "body." Because, let's face it, we aren't going to fully implement/host a full AI within the frame of a robot any time soon due to size and power constraints, at least not for a long while yet.
Argue that remote brain argument with my late cousin that was paralyzed from a neck injury. His brain was still in his body, yet he had zero control from the neck down. The human body ain't modular. Yet...
Thank you for asking that question! It's a great point. I hope that is the information you're looking for. (Scary that we can emulate AI responses so well - gah. off to go rewatch some Black Mirror!)
Upvote
-8
(0
/
-8)
Upvote
10
(10
/
0)
Humanity has been walking into this disaster for some time now. As someone else put it:
Upvote
26
(26
/
0)
Not fair! You're asking about things I learned 40-50 years ago. I hate pop quizzes!
Television? I remember the rudimentary parts from CRT's (The caps in flyback transformers HURT!) and the translation to LED/LCD, along with the old cameras using a specialized vacuum tube for image scanning and capture.
EM spectrum? When did EM merge with Spectrum? Okie fine, it is the full range electromagnetic radition like radio, microiwave, IR, regular ligh that we can see, UV, x-rays. and gamma rays, probably need to ask Hulk to be sure. Pretty sure there are 7...
Ohm's Law? V=I*R? Never could figure out why I stood for Amps /facepalm
Centimeter? Wavelength of krypton-80something times ? uhhh, 1.65k wavelengths? or was that a nm? Math is hard! I think I saw it changed since I was in scool to the distance light travels in 1/300,000,000 of a second or something. Need to look it up...
OMG, not pi! We had to learn it to 50 digits /whimper
Ask a hard one, like the difference between ionizing and non-ionizing radiation their effect on human tissue.
And these people are the ones making decisions for us and charting our futurre /cry
Upvote
4
(5
/
-1)
The situation with robotics is analagous to AI - quite potentially useful for specific tasks and environments but far from AGI or the robotic equivalent..
Upvote
9
(9
/
0)
jonbob_newcastle
Ars Scholae Palatinae
I daresay a squeeze bottle might be cheaper than an autonomous anthropoid robot but, if you must have a glass bottle, install a vibrator* in the hand.
Protip: when you buy canned goods like chilli or 'chunky' soups, store them on your shelves upside-down.
* No, not that kind. That one goes in the other hand.
Upvote
6
(7
/
-1)
I’ve seen more videos of theses robots failing than them being actually useful lol
Upvote
0
(0
/
0)
A machine that is working for a large portion of the day, every day, like a office copier, will also require constant upkeep maintenance. And my xerox isn't doing backflips.
Upvote
10
(10
/
0)
Upvote
4
(4
/
0)
