>”Don’t know what to tell you. Have you ever actually implemented a machine learning algorithm? Maybe go do one of the intro tutorials online. I actually use these algorithms for work.”

What you’re saying is at odds with everything I have read about this topic from every source. Pardon me if I don’t take the word of an internet stranger over everybody else in the field.

>”Why are cheaper goods a problem? Why is freeing human capital from assembly lines a problem? I really don’t understand why automation is a problem.”

“It is difficult to get a man to understand something, when his salary depends on his not understanding it.” -Upton Sinclair

Also your reply does not actually answer the question I asked and appears to simply be an attempted deflection. Cheaper goods are not a problem. People not having money to buy them is.

The core issue here which you’re doing some impressive contortions to avoid confronting is that there is no foreseeable barrier to AI improvement which will stop it from being able to automate every task which currently requires a human. At that stage, even if automation were to create a trillion new jobs, they’d all be possible to automate, so it would not solve the problem being discussed.

>”Considering those types of jobs yield higher fulfillment yes. I would far prefer a world where people can work on things they find interesting than forcing people to work mindless assembly line jobs.”

That wasn’t the question. I did not ask if you WANT everybody doing those jobs. I asked if our economy NEEDS that many people performing that class of labor. The real world alternative to shit jobs isn’t better jobs, it’s no jobs.

>”Not really. I think you know I’m talking about the industrial revolution but maybe this is a helpful point because a current continuation of AI is faster AI.”

AI doesn’t just “become faster”, it gains new capabilities. This is a failure of imagination on your part and an expression of your apparent belief that AI today has no capabilities that it didn’t have ten years ago.

>”I do this professionally. I own a business. I’ve worked in large businesses. That is what happens. Some specific examples, faster classification, better regressions results, automated parameter finding for models, less data munging, improved inventory allocation decisions, less manual data cleaning etc.”

What you’re saying is at odds with everything I have read about this topic from every source. Pardon me if I don’t take the word of an internet stranger over everybody else in the field.

>”It really is. Look at long-term (greater than 30 year data) and we see that world wide. There has been some stagnation in wages in the US. My personal guess is that globalization played a role in that. I also think at less than 5% unemployment you’ll start to see that problem getting fixed.”

What you’re saying is at odds with everything I have read about this topic from every source. Pardon me if I don’t take the word of an internet stranger over everybody else in the field.

>”Those jobs going away frees up humans to do better things. You’re coming across like Malthus or one of the classical economists who just can’t imagine how new ways of doing things could occur.”

Isn’t that you, earlier in this argument, saying that AI will never do anything that it can’t already?

>”I’ll go back to my previous point…maybe go implement one of these models?”

No, answer the question. When do you think AI will stop gaining new capabilities, and why? What barrier will prevent further improvement?

>”Ok. But you’re using a non-standard definition then. These techniques have been used to help solve operations problems since the 1940s. Why is this one application to a specific problem an indication that something is different. Are we now claiming that a hamiltonian, dynamic programming, or some other technique for minimizing or maximizing an objective function is something new? Its not.

I didn’t say it was new. We weren’t originally discussing whether it was new, but whether it constitutes creativity. You keep answering in ways which subtly shift the discussion and assign claims to me that I didn’t make. I notice when you do that.

There were no computers in the 1940s, so it should be pretty clear that the important development since then is that evolutionary algorithms can be done by computers. Evolutionary algorithms permit computers to generate useful new information without being conscious or conventionally intelligent. You say this isn’t how anybody defines creativity. However:

“Creativity: the use of the imagination or original ideas, especially in the production of an artistic work.


inventiveness, imagination, innovation, innovativeness, originality, individuality”

While the innovations being generated are not artistic, it satisfies every other part of this dictionary definition. It is producing original ideas and being inventive/innovative.

>”Fair enough. Just explain to me how one of the current trends in AI or machine learning is going to actually cause what you’re saying.”

Economic forces. The desire of the ownership class to reduce overhead expenses as much as possible will necessarily entail development of AI which replicates more and more human capabilities until there is nothing left humans can do which AI can’t.

>”I can see how these trends could lead to some levels of inequality we don’t like, but the current technology doesn’t iteratively improve to what you’re arguing for. It would have to jump to something new.”

Like it has many times before. Like I said earlier AI does not just “become faster”, it gains new capabilities. AI can do things today which it couldn’t just a few years ago. That’s not the result of improved processing power, it’s the result of software engineers improving their understanding of how to quantify and replicate human capabilities in software.

I post text here, often accompanied by images and sometimes video. People then clap or don't depending on whether they enjoy what I posted.