I saw a commercial the other day for a training school for careers in the trades-- plumbing, HVAC, electrical, etc. It showed various people saying things like “‘ChatGPT, fix my plumbing’…‘Sorry, I cannot do that’”. “‘Siri, get my furnace working’…‘does not compute’”.
It was an effective way to convey the message “you can have a well-paying job that’s future-proof”. But I started wondering, for how long? We’ve had a lot of articles about the rise of AI and how it will affect (or is already affecting) content creation jobs. But robotics is also advancing at an amazing pace-- we’ve all seen the videos of the dancing robots or the terrifying headless robot dogs.
I can think of a few reasons why robots may not routinely replace humans for hands-on trade jobs anytime soon:
Sure, robots can dance and do parkour, but the dexterity of the human hand, not to mention the intricacy of hand-eye coordination in performing close-up tasks, to which humans excel, is probably very difficult to replicate in a bot.
Even if and when that dexterity can be replicated, the cost of such a robot will likely make it impractical for all but the most specialized or dangerous tasks for a long time.
But I’m not really that knowledgeable of what the state of robotics + AI is at the present time. Maybe AI powered trade work robots are not that far off?
I suppose that depends on how well AI can think outside the box. A few years back, we had a very very slow draining sink. Drano didn’t work. Snaking it didn’t work. The plumber we called was stumped. Then my husband decided to work his way along the drain pipe from the basement side, taking down a couple small sections of wallboard. And that’s when he discovered that when we’d installed one of those boards, we’d shot a couple of nails right into the pipe. Gunk accumulated on the nails and all but blocked the pipe.
That rather sums it up, I think. An AI robot may soon be able to do routine jobs where the procedure is well understood and no complications arise. Though machine vision and co-ordination aren’t yet up to the level that LLMs seem to achieve for language.
If there are unexpected issues, though, you probably need strong general AI, and I don’t think we have a path to that yet.
For one thing, you don’t send C3PO to do your plumbing. You send some specialized plumbing robot that doesn’t have to lay on its back and swear as it tried to undo the rusted bolts under your sink. Which means that instead of five robo-fingers you have either specialized attachments or perhaps more general appendages that still don’t rely on human fingers. For example, something that could use an extending and highly articulated appendage with camera sensors on it to do soldering inside a space without having to cut a big hole in the drywall is worth more than an android in overalls.
Good example. I’d think that one of the first diagnostic things a robot would do when faced with a pipe blockage would be to extend a camera down the line and see where and what the blockage is.
We are decades away. A career away. Robots aren’t even in the ballpark yet.
One of the fascinating things about the alleged robot revolution companies like Tesla are promising is how poorly thought out their robots are, for just this reason. Why are they so humanoid? Well, I know why, because that’s what Elon Musk et al. see in science fiction movies. But that’s stupid; why should a robot be limited by the limits of the humanoid form?
Why can’t the Tesla robot telescope its legs and arms to it can reach 20 feet up? Why are its legs humanlike at all? Why don’t they bend the other way like a bird’s? Why don’t they have four or six appendages that are more articulated than a human arm? They’re stupid, and clearly just for hype.
Right. Androids are sexy (even the ones not designed to be sexy) and get the shareholders excited. Some tracked disc-bot looking for wiring faults isn’t and don’t. But one is a better idea for doing electrical work than the other.
This may be somewhat of a cultural stereotype, but I have often wondered why Japanese companies seem to have such a fascination with humanoid robots? Asimo, etc.
I agree with what others have said: a robot to do trade jobs is more likely to look like an industrial machine than a human. There is I suppose the argument that a lot of tools are designed to be used by a human hand…. but that sort of goes away once the tools are built in to the robot…
That said, I think that if the architects were to design buildings with robotic installation in mind, it’s not at all out of the realm of possibility that a lot of new build construction could be partially or fully robotic in the fairly near future.
Of course, this would require the same sorts of standardized and/or pre-assembled parts similar to the way that industrial robots work. So you’d have a cabinet that would be of specific dimensions and with the framing in specific spots, and the robot could come in, very accurately position Sink Assembly P23931 and connect the pipes. And the day before, a different robot would have run the pipes through the very specifically installed framing/joists and installed Kitchen Cabinet C138982 on the wall in the exact spot, so that Sink Assembly P23931 installs without any hitches.
I think if modern-day AI/robots are expected to deal with the level of slop that’s inherent to modern-day construction, or worse, old buildings where all bets are off, they very well may struggle with things not being plumb or where they’re 1/2" off from where they’re supposed to be, and so forth. There’s an inherent degree of creativity involved with being able to imagine what the problem might be and how to best solve it. So for @FairyChatMom’s nails through the pipe problem, even if it just ran a camera down and identified that there’s a blockage, present-day AI can’t make that leap to “That’s a nail” and extrapolate that it’s going to have to remove wall board and replace a section of pipe just because it sees a blockage. Not without an actual person training it to recognize that and solve it like that anyway.
One way to think about it is that we’ve had robotic automobile assembly for decades, but we still aren’t seeing robotic mechanics. And vehicles tend to be somewhat more dimensionally accurate and have less variation than old buildings with decades of sketchy DIY repairs.
Frankly, I don’t see this becoming feasible until trade utilities (and the structures they’re built into) are designed to be maintained by robotics.
Houses can be so eccentric and weirdly-built that even skilled human trades crafters are baffled. You’d need a genuine super-human general-purpose AI married to superior tools (like enhanced sensors) plus a body so flexible it can work from beneath the floor to above an extra-height ceiling.
And in all the purposes you can put such a robot to profitable use, would trades really be the most profitable? No. I think war would be far more money-making.
Therefore, you’ll never see fully sufficient carpenter or plumber or electrician bots because fully capable warbots will have already extincted the human race.
I think part of the reason for humanoid robots may be that they are more acceptable to the general public. Especially the robot prototypes meant for demo purposes. As long as they don’t get too humanoid looking and cross over into ‘Uncanny Valley’ territory. As robots become more acceptable in everyday life, their design may become freer to adapt to their function.
It’s similar to how computer interfaces and website design used to be more skeumorphic– that is, they looked like a digital representation of the actual thing they were copying. This helped people understand what the icons were for in the early days of WYSIWYG computer interfaces, but as people became more computer-savvy, interface design was freed up to become cleaner, more minimalist and abstracted.
I mean, factories are full of robots. They don’t look like people. Granted, they’re not generalists, but the human form is not ideal for a generalist, either.
This. Houses can be so different. My master bath has five fixtures (two sinks, toilet, tub & shower). One of the vanities has has childproof locks to access under the sink. It’s going to take some effort just to get to the proper location.
I had a guy in to fix my water heater recently. I mean, just negotiating a different space to get to my water heater, versus someone else’s basement or furnace room, involves an immense number of variables and physical problems. A human can do things like “not trip over that” and “don ‘t whack your head off a low occlusion” effortlessly, and that’s before you get to the ten thousand things that might be wrong with the damned thing.
I have no doubt someday robots will be able to do this but we are a long, long way off.
What about specialized plumbing robots that are operated remotely by cheaper labor overseas? That could happen well before you get general purpose plumbing robots or something.
My friend had a sign on the wall, “Beware of Bob” – Bob is their name for the mutant hockey puck that sometimes roams about the floor vacuuming up dirt, and taking about 5 times as long to do the job as a human would. Bob is a crude robot, but it shows a non-anthropomorphic form.
Some upper-end communities are highly exclusive and trying to steal “nature” from everyone else. As such, they tend to be not close to anywhere, which means that getting a tradesman out there can be problematic, as the nearest such person lives in a proper town, hours away. This is where tradesbots will start to show up first. They can sit in a community garage, being charged up, waiting for a service call while not offending the delicate sensibilities of the resident elites or displaying an unseemly ass crack while getting a job done.
.