Humanoid Or Specifically Built Robots VS Human Workers

[Spacescifi]

Consider a setting where humanoid robots with the same mechanical abilities as a human body are actually a reality.

What they lack is the ability to reason and rationalize like humans do. They simply follow whatever program given within specific perimeters.

I was curious... would such robots be a proper replacement for human labor jobs? Where much thinking is not required?

My thoughts: Yes, but they would need a human overseer or overseers to handle or oversee problems that come up that they are not programmed to handle. Like say the see a leaking roof? Report that. Say the item they are sent to look for is not where it is supposed to be. Report it.

Specific robots: The humanoid shape is a jack of all trades shape. It can build or design everything but it is not the most efficient at specific tasks. Specific robot bodies CANNOT build or design everything but are more efficient at a given task. Example? 

Spoiler

 

Instead of a group of guys with shovels.

 

What is most optimal?

If the environment is lethal (Mars or space) use all three. Namely humanoid robots to do specific human jobs to free up the crew or not endanger them.  Efficient specic robots where efficiency is key, and human overseers to assist robots when programming fails to solve something or it has a question.

Your thoughts?

Edited September 5, 2020 by Spacescifi

[magnemoe]

You can easy make an generalistic robot who is not human shaped. Main benefit with an human shaped robot is that it can directly use human equipment. 
However an robot using an modern wheeled loader could simply dock itself on the back of it and directly control it using the loaders cameras. 

Then done he can park it undock, pick up an shovel and do fine work.
You only need legs on an robot. in very rough terrain and stuff like stairs and more so ladders so you might want it on an ship but not in an factory. 
You can swap out bottom if needed, an pretty small footprint base in an factory to something with tracks on an construction site. 

[Shpaget]

It's gonna be a while before robots can solve random problems they were not programmed to solve, but it's trivial for robots to outperform humans at countless tasks, when properly programmed. That would require what is called artificial general intelligence (or strong AI), and it's not something we're close to cracking. Compared to strong AI, weak AI is simple and we've been working on it for a number of years. An example would be a neural network that can come up with a solution that it was never specifically programmed to find, but is in a very narrow field, like a chess game. There are so many possible chess games that it is not practical (or even possible) to program the computer to know the best move in each possible position, but there are techniques where you can show the computer a bunch of different games, and it just figures it out by itself.

Regarding industry, robots have been used in industrial settings for decades, because they are so much better at repetitive tasks. They are faster, stronger, more accurate, capable of working without breaks, and most importantly for industry, cheaper than humans. When they break down, you toss them aside and replace them with a new one. You don't need a humanoid robot for this. It would be way too expensive and unnecessarily complicated. In most cases a robotic arm with a couple of specific sensors and 6 degrees of motion is enough for a specific job. it would be a waste of resources to use a humanoid robot with all the joints and sensors a human has, if all you need is to transfer a box from one conveyor belt to another.

[DDE]

@Spacescifi, you forget Dmitry Rogozin's pesky little obsession. Human mind, robot body... why not both?

[kerbiloid]

1 hour ago, Spacescifi said:

What they lack is the ability to reason and rationalize like humans do. They simply follow whatever program given within specific perimeters.

I was curious... would such robots be a proper replacement for human labor jobs? Where much thinking is not required?

You just have described any modern corporation.

[Guest]

Industrial robots are a thing even today. The biggest problems they face today is mobility. No locomotion system has yet surpassed walking in terms of sheer flexibility. The problem is, walking is hard. Especially walking in very rough and/or unstable terrain, which requires you to think carefully about every step. Humans can do it, robots can't, because it turns out it would require enormous computational power for every step (that's another way human brain is awesome). Teleoperated machines can solve that, but in that case, you're getting the best trains of humans and robots at the combined price of both. Because of that, this approach is limited to situations which neither humans nor robots can handle by themselves.

In a setting such as a factory, humans can't beat robots when it comes to low-level tasks. Any workplace that's wheelchair-friendly is likewise a good place for robots. On the other hand, a robotic construction site would have to operate in a very particular way to make the most of its robotic machinery. The primary factor influencing the spread of robotics will always be price. A human construction worker is cheaper than a robotic one, so companies will hire humans. Once the reverse is true, for whatever reason, companies will start buying robots.

2 hours ago, kerbiloid said:

You just have described any modern corporation.

No, he only said thinking was not required. That way yesterday (90s-early 2000s). In today's corporations, it's outright persecuted.  The only reason they didn't go all-robotic is that cubicle open-office-plan monkeys are actually cheaper and more expandable. If they treated their robots like they do employees, the warranty wouldn't cover that.  

Edited September 5, 2020 by Guest

[Starman4308]

The primary limitation to multipurpose industrial robots isn't the form factor. It's the programming, and to a lesser extent, some types of sensor like smell and fine touch. Humanoid robots without the programming for multiple tasks is just a waste of money... and once you have the money to write the software for a new task, you may as well build a bespoke robot (not to mention fewer things can go wrong with single purpose equipment).

A humanoid or other general purpose form factor (e.g. spiderbot) really needs strong AI to kick off.

[magnemoe]

2 hours ago, kerbiloid said:

You just have described any modern corporation.

Or government as in any rule by large comity system works like this. However all the humans inside it are and have their own agendas. 

[Spacescifi]

2 hours ago, Dragon01 said:

Industrial robots are a thing even today. The biggest problems they face today is mobility. No locomotion system has yet surpassed walking in terms of sheer flexibility. The problem is, walking is hard. Especially walking in very rough and/or unstable terrain, which requires you to think carefully about every step. Humans can do it, robots can't, because it turns out it would require enormous computational power for every step (that's another way human brain is awesome). Teleoperated machines can solve that, but in that case, you're getting the best trains of humans and robots at the combined price of both. Because of that, this approach is limited to situations which neither humans nor robots can handle by themselves.

In a setting such as a factory, humans can't beat robots when it comes to low-level tasks. Any workplace that's wheelchair-friendly is likewise a good place for robots. On the other hand, a robotic construction site would have to operate in a very particular way to make the most of its robotic machinery. The primary factor influencing the spread of robotics will always be price. A human construction worker is cheaper than a robotic one, so companies will hire humans. Once the reverse is true, for whatever reason, companies will start buying robots.

No, he only said thinking was not required. That way yesterday (90s-early 2000s). In today's corporations, it's outright persecuted.  The only reason they didn't go all-robotic is that cubicle open-office-plan monkeys are actually cheaper and more expandable. If they treated their robots like they do employees, the warranty wouldn't cover that.  

 

I see. So the only way robots totally replace humans are either repetitive tasks without thought, which we already do... or human teleoperated robots, which we also use for military flying drones.

I think the reason mini tank land drones have not caught on is because as far as military is concerned... humans are cheaper.

In other words... when in war the preserving of life is not priceless, it has a price that sometimes is higher than those at the top are willing to pay. So instead of armies of mini tank land gun drones with smaller human squads, it is cheaper to risk life and limb of all human soldiers, augmented with equipment.

I suppose a high tech foe could jam telecommunications of the land mini tank drones.

But low level tech gun toting foes would not stand much chance against mini tank drones... without resorting to IED's anyway. Which they already use effectively against human soldiers.

Edited September 6, 2020 by Spacescifi

[SOXBLOX]

1 hour ago, Spacescifi said:

I think the reason mini tank land drones have not caught on is because as far as military is concerned... humans are cheaper.

Maybe they're cheaper in some militaries, but not Western ones. There aren't drone tanks mainly because        A) drones don't need armor, so they won't look like tanks, and B) only humans can currently discriminate between a dangerous suicide bomber and a harmless young man with a box of food.

Anyways, robots will likely not need to be humanoid. You might want, say, two tracks and manipulator arms above. Much easier, and nearly as versatile in artificial environments, like a space habitat.

In future, most jobs will revolve around finding things for robots to build or make, building the robots, or doing highly complex repair work. That will be cool, for the folks alive then.

[Spacescifi]

1 hour ago, SOXBLOX said:

Maybe they're cheaper in some militaries, but not Western ones. There aren't drone tanks mainly because        A) drones don't need armor, so they won't look like tanks, and B) only humans can currently discriminate between a dangerous suicide bomber and a harmless young man with a box of food.

 

Teleoperation means a drone COULD tell whether a man has a harmless box of food or a bomb. Like thermal and infrared vision to count just a few.

In fact, with proper sensor equipment, land drones teleoperated could easily be privy to information that a soldier's Mark I eyeball won't see.

Furthermore, teleoperation can be done via the human squad AND flying drones, suppossing satellites fail for some reason.

I said 'tank' for the tread tracks, not the tank shape per say.

Once weapons and sensor and telecom equipment is installed it's good as a human soldier or better... at destroying.

What it is not good at: 

Reconnassiance: Everybody WILL notice and try to destroy the land drone. Chances of getting away is low. You either fight your way out leaving carnage in your wake or lose an expensive land drone.

Edited September 6, 2020 by Spacescifi

[SOXBLOX]

1 hour ago, Spacescifi said:

 

Teleoperation means a drone COULD tell whether a man has a harmless box of food or a bomb. Like thermal and infrared vision to count just a few.

In fact, with proper sensor equipment, land drones teleoperated could easily be privy to information that a soldier's Mark I eyeball won't see.

Furthermore, teleoperation can be done via the human squad AND flying drones, suppossing satellites fail for some reason.

I said 'tank' for the tread tracks, not the tank shape per say.

Once weapons and sensor and telecom equipment is installed it's good as a human soldier or better... at destroying.

What it is not good at: 

Reconnassiance: Everybody WILL notice and try to destroy the land drone. Chances of getting away is low. You either fight your way out leaving carnage in your wake or lose an expensive land drone.

Thermal=infrared

Tanks already have advanced sensors and computers to process the data for the soldiers. No Mk I eyeball needed.

Teleoperation can be jammed, might not work in urban environments or prohibitive terrain, and leaves the system open to electronic attacks other than jamming. If you put humans nearby to teleoperate the "tank" then they need to be protected by something. Armor boxes work nicely. Now how about an armor box on highly mobile treads, and with a gun on top to defend itself? Oh, wait, we've got that already...

[kerbiloid]

For human-compatible workplaces use human-like robots.
For others - don't use them.

Isn't that obvious?

[Spacescifi]

1 hour ago, SOXBLOX said:

Thermal=infrared

Tanks already have advanced sensors and computers to process the data for the soldiers. No Mk I eyeball needed.

Teleoperation can be jammed, might not work in urban environments or prohibitive terrain, and leaves the system open to electronic attacks other than jamming. If you put humans nearby to teleoperate the "tank" then they need to be protected by something. Armor boxes work nicely. Now how about an armor box on highly mobile treads, and with a gun on top to defend itself? Oh, wait, we've got that already...

 

Wow.

So essentially land warfare just became like sea warfare, with the carriers being carrier tanks, and the command human tanks being the 'battleships', while the army of mini tank drones are the advance force that screen attacks directed at tank.

For this to work they would need a swarm of drone aircraft that can hover and shoot for extra protection.

[Guest]

1 hour ago, SOXBLOX said:

Teleoperation can be jammed, might not work in urban environments or prohibitive terrain, and leaves the system open to electronic attacks other than jamming. If you put humans nearby to teleoperate the "tank" then they need to be protected by something. Armor boxes work nicely. Now how about an armor box on highly mobile treads, and with a gun on top to defend itself? Oh, wait, we've got that already...

Well, if they're teleoperating another vehicle, then the command unit could dispense with the gun. Maybe have an MG/grenade launcher turret for defense, plus an APS for missile protection. In fact, you're looking at something more in vein of a heavy APC, and since you don't need the big gun, you might as well get several crews into one box.

In fact, this idea is not new, it dates back to WWII. Never again say that Russians can't do advanced technology  :
https://en.wikipedia.org/wiki/Teletank
It was just too expensive to deploy widely, but it worked. These were controlled from actual tanks, for tactical reasons (and to prevent capture, the command tank was to shoot the remote one if the enemy got his hands on it), and they were used in combat. The problems with that idea are mostly about development cost (it's easier to invest in incremental upgrades to gear dating back to the 80s or earlier), and to a lesser extent, politics.

[DDE]

7 hours ago, Spacescifi said:

I think the reason mini tank land drones have not caught on is because as far as military is concerned... humans are cheaper.

Hold up, I have an (unsourced) tweet about that:

 

2 hours ago, SOXBLOX said:

Teleoperation can be jammed, might not work in urban environments or prohibitive terrain, and leaves the system open to electronic attacks other than jamming.

That is the correct response. The Russian military went all-in on UGVs in Syria, and found the practical control range to be disappointing.

1 hour ago, kerbiloid said:

For human-compatible workplaces use human-like robots.
For others - don't use them.

Isn't that obvious?

I think we can skimp on leg day even in human-compatible workplaces.

Did I just uncover a conspiracy between legless robots and wheelchair-bound humans?

47 minutes ago, Spacescifi said:

So essentially land warfare just became like sea warfare, with the carriers being carrier tanks, and the command human tanks being the 'battleships', while the army of mini tank drones are the advance force that screen attacks directed at tank.

For this to work they would need a swarm of drone aircraft that can hover and shoot for extra protection.

It's quite possible we're headed there, yes. However, the tank drones can be... Well, not "mini". The Russian military has just tested the T-14 in unmanned mode, and have been doing the same to T-72 variants for several years. Back when they started with Uran-9 they were pretty clear they wanted unmanned battalions.

28 minutes ago, Dragon01 said:

In fact, you're looking at something more in vein of a heavy APC, and since you don't need the big gun, you might as well get several crews into one box.

I'll just park this here, then:

Edited September 6, 2020 by DDE

[kerbiloid]

3 hours ago, DDE said:

:

 

Nice to see that somewhere the kickbacks are higher than in Russia, affecting the total cost. Otherwise I couldn''t believe this table.
(And I'm not about Australia, just in case).

3 hours ago, DDE said:

Did I just uncover a conspiracy between legless robots and wheelchair-bound humans?

I would say that as the "green powah" prepares the humanity to the totally nuclear energetics by the mid-XXI (by plugging everything into outlets rather than pouring petrol in tanks),
so the "social availability", "friendly urban landscape", and so on, prepare the urban landscape for the robotization.
 

3 hours ago, DDE said:

The Russian military went all-in on UGVs in Syria, and found the practical control range to be disappointing.

Happily this is absolutely not about the uncrewed turrets on the combat machines where the crew seats in a capsule together with driver and aims the cannon through the cameras on the turret.

(I.e. teleoperates the turret.)

(Just recalls MBT-70 again).

Edited September 6, 2020 by kerbiloid

[kerbiloid]

3 hours ago, DDE said:

unmanned battalions.

Spoiler

 

[DDE]

2 hours ago, kerbiloid said:

  Reveal hidden contents

 

*Sings God Save the Tsar angrily*

Spoiler

Ok, ok, ok, nyash-myash, I will include you and your fetish.

Spoiler

 

[Terwin]

People seem to be forgetting: teleoperated is the 'pay for both' option.

In places where there is a high chance of loss of life(bomb disposal, scouting, offensive combat operations) we already have teleoperated drones(plus tanks and lots of missiles).

In places where it is more about interacting with people with very little risk(guard stations, check-points, security forces, training local militia), we use humans.

While I admit that I do not often watch the news, my understanding is that a large percentage of deaths in the US armed forces in recent years are from things like IEDs either along roads or used by suicide bombers at check-points/guard stations.  As opposed to most of the offensive actions which have been undertaken by drones(mostly flying drones), generally with operators in a secure location(only take-off and landing handled by local operators, the other, less ping-sensitive operations, handled by people well outside of harms way).

 

When there is not a risk of being attacked, but there is a need for robotic operation with human oversight, we generally call that 'heavy equipment' (cranes, tractors, bulldozers, back-hoes, excavators, dump-trucks, trailer trucks, etc) and we have been using those for >100 years(first bulldozer: 1923, first trailer truck:1896).

 

There is no need for robots in an office environment(aside from deliveries), as those are places specifically for human brain-work.  The robot brain-work is generally handled in places called 'data-centers' and are easy to access either from home or cubicle, so there is not much need to have them trundling around an office, as they can just be kept in a secure room(often called the 'Server Room') if you want them local.

 

TLDR: we are already doing all of this, just much more efficiently than you are suggesting, and very rarely does it make sense for a robot to make use of generalist legs when much more task efficient propulsion is available, and even then only in the cases where the 'robots' can't just remain stationary and positioned for easy maintenance.

Edited September 6, 2020 by Terwin
18-wheeler analogs are older than bulldozers and also fit 'local operated robot'

[kerbiloid]

Spoiler

  

2 hours ago, DDE said:

Ok, ok, ok, nyash-myash, I will include you and your fetish.

The fetish is not included, as the last image is not visible. But I get yor idea from ny-my, and it's obviously wrong.

If make anthropomorphic robots, we could use them as oarsmen for seaships.

Spoiler

Starting countdown to the robocatgirls.

 

Edited September 6, 2020 by kerbiloid

[SOXBLOX]

52 minutes ago, Terwin said:

People seem to be forgetting: teleoperated is the 'pay for both' option.

In places where there is a high chance of loss of life(bomb disposal, scouting, offensive combat operations) we already have teleoperated drones(plus tanks and lots of missiles).

In places where it is more about interacting with people with very little risk(guard stations, check-points, security forces, training local militia), we use humans.

While I admit that I do not often watch the news, my understanding is that a large percentage of deaths in the US armed forces in recent years are from things like IEDs either along roads or used by suicide bombers at check-points/guard stations.  As opposed to most of the offensive actions which have been undertaken by drones(mostly flying drones), generally with operators in a secure location(only take-off and landing handled by local operators, the other, less ping-sensitive operations, handled by people well outside of harms way).

When there is not a risk of being attacked, but there is a need for robotic operation with human oversight, we generally call that 'heavy equipment' (cranes, tractors, bulldozers, back-hoes, excavators, dump-trucks, trailer trucks, etc) and we have been using those for >100 years(first bulldozer: 1923, first trailer truck:1896).

There is no need for robots in an office environment(aside from deliveries), as those are places specifically for human brain-work.  The robot brain-work is generally handled in places called 'data-centers' and are easy to access either from home or cubicle, so there is not much need to have them trundling around an office, as they can just be kept in a secure room(often called the 'Server Room') if you want them local.

TLDR: we are already doing all of this, just much more efficiently than you are suggesting, and very rarely does it make sense for a robot to make use of generalist legs when much more task efficient propulsion is available, and even then only in the cases where the 'robots' can't just remain stationary and positioned for easy maintenance.

Most old SF books or what-have-you's include robots like this. If they could provide an economic benefit, they would have been built and deployed into the workforce in large numbers long ago. Of course, SF usually gets its predictions only sorta correct. Like you said, we have computers aiding us, they just don't look like people.

10 hours ago, Spacescifi said:

 

Wow.

So essentially land warfare just became like sea warfare, with the carriers being carrier tanks, and the command human tanks being the 'battleships', while the army of mini tank drones are the advance force that screen attacks directed at tank.

For this to work they would need a swarm of drone aircraft that can hover and shoot for extra protection.

Eh. Maybe, but it seems to be going in the direction of either "no more tanks" or "tanks have active defenses", like Israel's Trophy or Russia's Arena. Airborne drones are being launched from tanks to scout ahead, but these are only a few inches long, and aren't weaponized.

Much better than small drone tanks would be active defense-equipped (manned) tanks with tons of CAS from UAV platforms.

Edited September 6, 2020 by SOXBLOX

[magnemoe]

1 hour ago, SOXBLOX said:

Most old SF books or what-have-you's include robots like this. If they could provide an economic benefit, they would have been built and deployed into the workforce in large numbers long ago. Of course, SF usually gets its predictions only sorta correct. Like you said, we have computers aiding us, they just don't look like people.

Eh. Maybe, but it seems to be going in the direction of either "no more tanks" or "tanks have active defenses", like Israel's Trophy or Russia's Arena. Airborne drones are being launched from tanks to scout ahead, but these are only a few inches long, and aren't weaponized.

Much better than small drone tanks would be active defense-equipped (manned) tanks with tons of CAS from UAV platforms.

An humanoid robot is hard to make and is high maintenance, also most useful for stuff like repair who require a lots of thinking compared to assembly parts. 

Yes tanks might have UAV but it will not be carriers more like the spotting planes carried on battleships and cruisers between WW1 and 2. 
And not probably on the tanks but on the AFV. Tanks will get lots more active defenses 

Unmanned ground vertices will be important for guard duty. Probably fire support. Recon is another field, yes you could land an AUV but something with an sniper rifle might be better. 

[WestAir]

I am not an expert in any of the discussed fields,

That stated, in my uneducated opinion, I can see a mechanized workforce becoming the norm - especially in response to the fragility of service based economies as a whole. You already have automated checkout scanners, automated phone lines, automatic trains. We're working on automated cars and buses and planes.There will come a time, probably in our life-times, when any occupation you can think of can be done better by a machine; up to and including public office.

Where does that leave us? Better off, I hope. As for large militaries using mechanized assets to fight each other, with super-intelligent computers out-strategizing each other, I can't say where we fit in the mix. It'll certainly be interesting to see.

[Guest]

Computers won't be "super-intelligent" or doing any strategizing any time soon. They're still 100% deterministic machines, with no capability for thinking whatsoever, only computing. Mechanized workforce ends where intelligence begins. So you're right about public offices. Military officers (except maybe 2nd lieutenants ) and scientists should be fine, though.

I'd expect automated buses to still have a person onboard, if only to oversee the passengers. Automated cars will likely be required to have manual overrides (even if it's just an emergency stop), just in case of something they weren't designed to handle. If not from the outset, then the moment the first accident occurs which would have been prevented by such a system. The fundamental problem with automated systems is that they're great at following rules to the letter, and if that's all that's required they work well, but the moment the best course of action is to break the rules, they fail miserably. As much as the authorities like to pretend otherwise, rules are never perfect, and it takes a human brain to recognize such moments. It's also good when the judge who decided whether breaking the rules was justified has a brain, too, but that's not always assured (what with them all being lawyers ).