Comment by caseymarquis | Hacker News Reader

Comment by caseymarquis | Hacker News Reader

Globally. The title is somewhat misleading. That's roughly 0.4% of global jobs in 2019 being lost over a decade; So, something like 0.04% per year. I think society will survive.

If anything, those numbers are underwhelming. Having worked with industrial robots it's not surprising that they won't be taking over the world any time soon.

Robots are a large up front investment in a company's immediate processes. This investment is only good if the processes remain mostly static for a very long time. Even then, there is a major cost when processes finally change. It's stupidly similar to software development in so many ways.

Mass automation already happened, at least in my country (USA). The data shown in the article seems to indicate most of the increased use will be in countries like China. Rising wages are likely the dominant factor driving increased adoption in that situation, not new technology.

Everyone I know who actually buys robots views them as a trade off, not a panacea. As robots' capabilities slowly improve, their use will increase a bit, but operator guided machinery (ie CNCs, injection molding, task specific machines for things like packaging) glued together with flexible human workers isn't going anywhere until we hit something resembling AGI.

Can't speak to the use of robots outside of manufacturing, but that's not what the article is about.

Curious your thoughts on the upcoming wave of low cost “co-bots”. While still too expensive to make the analogy of the PC vs the mainframe, they have started approaching the 15-30k/base price. While obviously less capable, much like the PC, I believe they will have effective uses.

Thoughts on how this might change the equation?

In the larger scale type of manufacturing I'm working with, those small bots tend to have trade-offs that mean they're not really worth choosing over a larger, more traditional manufacturing robot, or a solution without robotics at all. Mounting difficulties, lack of range of movement, and lack of robustness being the core issues. In most scenarios, you're better off avoiding robots and going for something statically mounted. Failing that, choosing a Kawasaki model that's been in production the last 30 years. Then you know beyond a shadow of a doubt that it's going to run for the next decade without a hitch. You can find operators with experience relatively cheap. Every imaginable mounting head exists and has been in production long enough to know the issues you're going to face with it.

The big robot will survive in large scale manufacturing for the same reason Java survives in enterprise; it's part of the culture at this point. It does its job well, it hardly ever fails, and it's relatively easy to bring someone up to speed on working with them. Like Java, you'll probably see a number of companies pop up that prefer a newer technology, because it fits their use case well. But I'd expect the bulk of total money spent to continue to be on the kinds of industrial robots we work with now.

The place I can see small robots making an impact is knowledge work, such as laboratories or design. And realistically, in those fields we've never been time constrained. It's more common, in my experience, that the limiting factor is the number of good ideas a person can come up with. There's no way to reduce "hours formulating an idea" short of AGI. Scientists have to read papers, designers have to study a company or product to make all of the branding coherent.

All of that said, the newer, low cost bots will probably move the needle, but they won't accelerate the pace of automation significantly. More likely, they'll contribute to maintaining the current acceleration.

> but operator guided machinery (ie CNCs, injection molding, task specific machines for things like packaging) glued together with flexible human workers isn't going anywhere

There are funded companies already trying to get rid of all of that. Here's one - https://www.technologyreview.com/s/613743/relativity-space-3...

Give it a few years. It's going to happen.

Reply on news.ycombinator.com