People have automation backwards - automation doesn't solve problems. In order to automate, you have to solve all the problems. The machine has to know how to detect and handle any variation it encounters - or summon human help. If it has to summon human help too often, or fails to detect a problem too often, it is simply not worth it to automate.

I run into this all the time in my day job with software automation. Everyone thinks you wave some kind of magic wand when you say 'automate' that makes it just work. I have to put out the speed brakes once in a while when chatting with my VP to remind him that automation doesn't necessarily mean accurate, as much as it means fast. Sure, I can provision a bunch of network gear quickly, but if the automation is wrong, that just means I'll wipe out an entire portion of the infrastructure in seconds.

Added to https://github.com/globalcitizen/taoup

Neat! I've never been quotable before!

Typically not worth the cost of automation. Automated tasks need parts with certain levels of precision, among other requirements. That precision can some times make parts excessively expensive, heavy, etc etc.

One example of this that you don't see mentioned often, is that stamped door assemblies frequently need "adjustment", aka bending and beating, in to position to make them fit properly. You can see some examples of it in this video: https://youtu.be/iPhwBUnR_XA?t=216

Huh, fascinating! I always imagined that, if adjustments were needed, they'd be done on some sort of off-line jig where deflection could be precisely controlled. But no, it's just some guy using his body weight to bend the door into shape.

I've seen mostly 3-4 kinds of automation:

- Material handling - moving a part from A to B. Many of the parts looked too heavy for a human to lift. Some could probably be lifted by two humans working together, but that's a really bad use for humans and easy enough to automate.

- The presses and possibly some kind of moulding/pouring process, which obviously need to be done by a machine

- Spot (?) welding

- The coating/painting

So I'd turn the question around and say: It seems like they automated only a few key processes, and many of them because human labor isn't really an option. The presses replace many different production techniques, some of which could probably be done manually, so replacing those other techniques with "just put it in the press" is an advancement and it implies automation. The painting could be done by humans but the overall process would again be a lot more complicated. Automation for the welding is well established, and you want the consistency of robots.

> The painting could be done by humans

When was the last time a car manufacturer used humans to spray paint? Decades?

Funny you should ask that, I was just watching a video about car door manufacturing.

Here's the timestamped link showing a worker in a modern factory spray painting a door. https://youtu.be/iPhwBUnR_XA?t=442

Now, how common is that? No idea.

It's a pretty well-know story that Tesla's model 3, their first mass production car, was almost fully automated. It ended up making a ton of problems and Tesla moved some of the production to humans. It seems machines are very good at certain things, but not everything. Yet.

It's basically too hard to automate, yes. These may be mundane sounding tasks like checking if a screw is properly secure, we're still far from full automation, the robots simply do preprogrammed tasks, they have no sentience or intelligence.

Checking things like bolt tightness is a hybrid automation. There are machines which know where all the bolts are within a given area, and can test each one and adjust as necessary to make sure they are torqued. But that machine is itself handled by a human that gets it to the right spot.

I was kind of worried. Maybe Tesla had a general AI but kept it to themselves?

Elon mentioned [0] it is easy for a human to grab a dangling cable. The cable may not always be in the exact same place on each car, so you need a robot with a vision system.

[0] https://www.youtube.com/watch?v=MevKTPN4ozw

I can also imagine that it minimises the tail-risk, because a human can come and fix it. If he would not be there and a critical error occurs, it could take down the whole factory in time before help arrives (maybe even something trivial as a misplaced item on the belt).

Probably lower on the cost/benefit ratio of automating, or something that requires a decision to be made.

Isn't a common complaint with Tesla now their build quality?

The more automation, they more you need humans to check if it works.