People Who Train Robots to Do Their Own Jobs

> you all get a pension at 33% wages as long as the company is in business and 6 months to find a new job

Boss: "Or, and take a minute to appreciate my cleverness before you respond, how about zero, and don't let the door hit you on the way out!"

These improvements eventually materialize regardless of any incentives for this behavior. Perhaps it would come marginally sooner with an incentive.

But since non-owner workers will no longer have a relationship where they provide continued value to their employer, the company is (correctly?) motivated to screw them over (think single vs. repeated prisoner's dilemma). The company could even pretend to offer the incentive to get these improvements sooner, and then fire them anyway.

The game theoretic optimum for capital owners is to screw over their (soon-to-be former) employees. This is why some suggest a tax on automation -- there's a market externality here that isn't going to magically go away.

In most successful companies, the guy who came up with the idea and his manager will be heavily rewarded, and taken care of in a meaningful way. Everyone else is given some time to either find a different team to work for, or asked to leave and find another job where they are actually contributing something of value.

Although this might seem tragic in the immediate aftermath, it's precisely activities like this which lead to GDP growth and economic prosperity.

The biggest gains of course still go to the business owners and executives. Not the working class. Hence why we need higher taxes on the wealthy and a stronger social safety net.

Companies don't make decisions. People do.

I hear this often, but have yet to see it in practice. I have worked in government and a number of underfunded non-profits who would dearly love to have some of these things automated. But in my experience it just shifts work. It automates something, requiring extra work somewhere else. Or it automates, but then breaks one, mission critical function that needs to be taken care of. I truly don't think this is just about "save my job", because they feel the pinch of lack of personal and funding, but about the difficulty in truly enumerating what the workflow actually is. Once you've paid a consultant to take care of it, you realize you forgot 3 steps in the process or that they were not listening when you described them.

I guess. But it strikes me as engineers and coders will just gain more power to work in higher-level abstractions.

I think this happens all the time. Since good programmers can automate practically everything, they tend to automate as much as makes economically sense for them.

So programming, almost by definition, always involved all tasks that are not automated yet. Programming gets easier and harder that way at the same time: Previous tasks (e.g. taking care for memory management) become easier or automated way, while harder tasks remain (larger, more complex applications, higher abstraction levels).

So I'd rephrase it that way: Engineers and programmers are effected every day by this "threat", but so far worked quite well for to their benefit, they just operate mostly on higher and higher levels, while still having to keep in mind (and having to fix) the lower levels.

Kind of full stack, with a growing stack, thinner and thinner on lower levels, but overall higher and higher.

I disagree. Programming is the act of interacting with AI (the compiler) in language (code) detailed enough to specify what we want it to build for us. As tasks become abstracted to a single click or line of code, the programmer is free to work at higher levels of abstraction.

Most (e.g. [1]) don't think of their job as being automated because most don't understand how it will happen. The issue is we only have the present way to think about how the future will work. The future will likely eliminate a lot of mundane engineering jobs (e.g. CRUD apps) with tools that provide automation (learned or programmed) and more intuitive interfaces for the layman. Not all work can be automated, but those are the edge cases. The focus is to eliminate most work, by building smarter and more niche services.

For example, Squarespace, Wix, Shopify, et al have reduced the number of jobs for CMS and ecommerce developers. Zapier and IFTTT have reduced the number of integration jobs. The Grid was an experiment in creating whole websites automatically (most regard it as a failure, but the tech will likely catch up). There's research into using ML to identify and correct bugs in code. These are the low hanging fruit that will erode away at the jobs easiest to automate, leaving behind more niche work.

It's hard to imagine less work for engineers in the future because there's so much work to go around right now. That's likely the case presently because there's so many industries where manual labor can be eliminated, creating lots of new labor for engineers. Whole industries may be disrupted until corporations with automation emerge and take over. We'll likely see engineering work keep rising until there isn't as much left to automate and engineering supply catches up, at which point we'll hit "peak software engineering."

There will be engineers in the future just like there will be medical professionals in the future, but they'll probably be doing a different job and there will be a lot less work to go around. I like to think of that line of work as "escalation jobs," where when the machine fails, it goes to the human. Other work will likely get more and more scientific, theoretical and research-oriented.

1. http://www.cnbc.com/2017/02/24/americans-think-robots-will-t...

That is literally two years or less away from today. Please look up momentum machines. They're beginning to hire people from the food industry and have stopped hiring people from the tech field vs a couple of years ago. It's getting close.

42

That's the way I see it. Programmers have been automating themselves out of menial tasks since the first compiler, or possibly even before.

What is considered "menial" depends on the context of the problem needing to be solved and a mix of your personal experience.

The increase of hardware performance also helped allow software abstractions to the ridiculous levels you see. That's what also makes it possible for computers to solve a greater array of business problems.

Engineers making tools that are used for making other tools, that are tools for other tools for other tools that eventually helped make someone's wedding photography site on Wix.

I'm guessing that's not a realistic option for people in jobs where automation is taking place. The companies that would otherwise hire them are probably automating or planning to automate.

If they have the resources, then retraining themselves for another job might be worthwhile, but just leaving their current job probably won't help them much.

Severance payouts

Your severance package is dependent on your training your replacement. We're talking months worth of pay which is nothing to sneeze at since you have nothing to gain from walking away from that severance.

Besides severance, there's unemployment which you can't claim if you quit.

In theory the purpose of automation is humanitarian. In practice results vary.

If I truly believed that automation, especially in the short term, would include UBI or some other social safety nets, I'd be far more optimistic about this.

I'm nervous that in the short term, fewer people will continue to get very rich, and more people will fall out of the middle class.

Is it because we are not ready to do away with capitalism as it becomes obsolete?

> we as a society don't understand that the purpose of automation is humanitarian.

The purpose of automation is humanitarian in the same way that politicians serve the needs of their constituents, and police protect their communities.

While I reap the positive benefits of all of the above, there are plenty of marginalized people who are harmed by exactly the same forces.

In the gap between theory and reality, there's a lot of misery.

> we as a society don't understand that the purpose of automation is humanitarian

If "we as a society" don't understand the supposed purpose of something that we as a society are enacting, then that supposed purpose is academic and ultimately meaningless. There is no universal agency or purpose behind automation other than that of those who are building it, who to date have largely been capitalists and not utopians.

Fair point. Perhaps "perpetual" is a better word.

Right, but not started on a good footing for sure, kind of like if you are born into abuse and your epigenetics are all peaked on stress expression, harder to integrate harmoniously into positive environments. Speculation vacation... ok I'm back.