Ted Chiang: Elon Musk's fear of runaway AI is a projection of his repressed terror of runaway corporations


Originally published at: https://boingboing.net/2017/12/18/skynet-llc.html


Let me take you down
'Cause I’m going to strawberry fields
Nothing is real
And nothing to get hung about
Strawberry fields forever


So the problem with AI is that it will be programmed by amoral capitalists? To think that we USED to worry that AI would be programmed by the military.


Maybe if we relabel corporations “hive minds” they sound scary enough to do something about.


The time to worry about AI is before they exist, not after (at which point it may be too late). There is good precedent in the warnings about nuclear weapons from numerous scientists before they existed. No, nukes were not “just another bomb”, but became an existential threat.


The point is that they already exist, except that they are carried out mostly by human operators rather than by computer processors.


How about “Move fast and break everybody else’s things”? That seems to be the true principle behind all these “disrupters” like Uber, and so on, and indeed seems to be a basic capitalistic strategy, looking back.


That particular Venn Diagram is become more and more of a circle, which doesn’t help either.


Runaway capitalism is frightening. Runaway AI is frightening. Runaway AI built by immoral/amoral runaway capitalistic entities is… frightening squared?


Corporations are people, my friend. Psychopathic people, mind you…


Another in the same vein, although it seems a bit weak.



This is the problem we are mired in and can’t even see.


I read that one, too. He misses the key point, which is that if Romney and the Citizens United crowd are going to treat corporate personhood as more than a legal fiction then it’s fair game to apply Hare’s test (which categorises even the “saint” companies as sociopaths, albeit ones wrestling with some success against their own nature).


Corporations are indeed conscious entities in my opinion. Not all of them are psychopathic – the one I work for (and have helped to build) is pretty cool. They certainly are byproducts of their owners and workers, to be sure. They should also be held to the legal and ethical standards of humans, I think. They are what we make them, and what we allow them to express as. Our inability to properly control them speaks mostly to our own weaknesses. I do not feel that those weaknesses are insurmountable.


Conscious entities, yes. Human (as in individual people), no. That’s why Chiang is so on-point in his parallel with AIs.

I’ve built companies, too. At no time did I fool myself into thinking that their personhood was anything more than an expedient legal construct. Accepting Romney’s alternative would carry the depressing implication that my job was trying to control a psychopathic moron (I leave that to the White House staffers currently trying to do that with the human embodiment of his self-named corporate entity).


This is such a weirdly backward analysis.

Musk is perfectly well aware of the defects of corporate captitalism.

There are good reasons why SpaceX is still a privately held company, with no shareholders to answer to. (He would have preferred to keep Tesla private as well, but it required far more capital than could be raised privately.)

Knowing full well the dangers of (primitive, human-speed-limited) artificial corporate ‘hive minds’, he is understandably concerned* about what could happen if (human-smart+ and way, WAAAY faster) AI gets hold of those same corporate levers.

It’s not Musk’s job to Fix All The Things Everywhere. Uncontrolled corporate power is a problem, but it’s a problem that already exists, and is already deeply embedded in our culture.

It’s not something one billionaire can fix with a research project.

But one billionaire might be able to head off some of the more… umm… unexpected of the tech’s (inevitable) unintended consequences.

One of the ways he’s doing that is spending millions on a non-profit dedicated to building an open-source AI, so that AI can’t become the exclusive domain of any corporate oligarchy.`

Another way is spending yet more millions to develop a direct human-machine neural-link interface, in hopes it might improve our odds of keeping AI benign and beneficial

Yeah, I know. Musk is spending his money on his own concerns rather than on yours. That seems to be a popular complaint.

*Please note that “understandable concern” <> “fear”.

I’m appalled that most tech analysts can’t seem to see any middle ground between “starry-eyed uncritical enthusiasm” and “gibbering irrational fear.”

Anyone who thinks this is “fear” talking is really not paying attention to what’s actually going on.


Sometimes wonder if it isn’t the more you know about science the scarier it gets. I found that story about self learning ai programs coming up a humanly incomprehensible language when nobody had thought to put in rules about it pretty weird. Studied a bit of genetics and started wondering if I should start worshiping a god called DNA.


Study physics and information theory (feel free to throw in some occultism to boot), and you will realize that “god” is an AI.


Yes, that’s a bogus analysis crafted to make a social and political statement.
A point that is to a big extent valid, but that has jack and squat to do with the problems that AI rises.


Amoral captitalists in the military-industrialist complex.
There, synthesis complete!