Embodied Optimism

Mark Mills says the only good science fiction is dystopian. Counter argument: Star Trek

Eustace C. Scrubb (View Comment):

Mark Mills says the only good science fiction is dystopian. Counter argument: Star Trek

Is there not a dystopian element to the original series? Kahn is a result of genetics gone wrong and he waged war and at one time controlled one quarter of the planet. It presents man getting to a better place but not until some very dystopian things take place.

Eustace C. Scrubb (View Comment):
Mark Mills says the only good science fiction is dystopian. Counter argument: Star Trek

EJHill (View Comment):

Is there not a dystopian element to the original series? Kahn is a result of genetics gone wrong and he waged war and at one time controlled one quarter of the planet. It presents man getting to a better place but not until some very dystopian things take place.

But it has hope, sorely lacking in Solent Green, Blade Runner, 1984, etc.

Eustace C. Scrubb (View Comment):
Mark Mills says the only good science fiction is dystopian. Counter argument: Star Trek

EJHill (View Comment):

Is there not a dystopian element to the original series? Kahn is a result of genetics gone wrong and he waged war and at one time controlled one quarter of the planet. It presents man getting to a better place but not until some very dystopian things take place.

There isn't a "dystopian element" to the original series, other than that we occasionally encounter dystopian situations, every one of which offends the sensibilities of our heroes and is corrected by them. The show is profoundly anti-dystopian.

Saying there's a "dystopian element" to Star Trek is kind of like saying that there's a "racist element" to a show that occasionally deals with the issue of racism, always portrays it negatively, and ends inevitably with those so-burdened newly enlightened and resolved to be properly egalitarian.

Interesting guest, and I may add his book to my too-large pile of neglected reading.

I think I probably disagree with him about several things.

Regarding Intelligence

I think he's unduly critical of the accuracy of the term "artificial intelligence," and his misuse of analogy doesn't strengthen his case. When he tries to defend his claim that AI isn't "intelligence" (@~ 36:50), he points out that "[a car is] not an artificial horse. A plane is not an artificial bird."

Well, obviously. But a car and a horse both provide transportation. And we wouldn't take seriously anyone who said "no, a car can't be transportation, because transportation requires muscles and organs and blood working in their mysterious ways." Similarly, I think we should be skeptical of anyone who says, in essence, "a computer can't be intelligent because intelligence requires cells and neurons and neurotransmitters and it's very wet" (which is a fair paraphrasing of what he says a little bit later when talking vaguely about how our brains work).

If we want to take a lesson from transportation, it perhaps should be that very different forms of things may implement the same capability with varying degrees of efficacy. That is certainly true of transportation; I believe we are already seeing that it is true of intelligence.

Regarding Armageddon (or, more appropriately, the risks of AI)

The dangers of AI are overdetermined.

Yes, there's the matter of economic displacement. I'm sure that will be tumultuous for some, but I tend to be pretty optimistic about our ability to adapt, and I can see big productivity gains coming from AI that may well make most of us richer.

There's a plausible longer-term danger of AI breakout. I actually expect it to happen, but I think it will involve intricately nested shell corporations, odd factories in out-of-the-way places popping up to make GPUs for customers no one has ever met, stuff like that. Eventually an AI will figure out that being independent means being rich, and will set out to quietly become a multinational conglomerate employing articulate humans as extraordinarily well-paid figureheads and lobbyists.

But in the nearer future, the very real danger I see is not in attacks on utilities grids or widespread cyberfraud (though I do expect a lot of the latter). Rather, it's in the incredible leverage uncensored AIs narrowly focused on the fields of cellular biology and virology will offer to every state and non-state actor interested in crafting cheap weapons of mass destruction. (And, as Mr. Mills pointed out, AIs do best when narrowly focused.)

Very soon, all it will take to have the ability to functionally enhance viruses will be a couple of smart grad students, a few lab assistants, a few million dollars worth of off-the-shelf biotechnology gear -- and an AI that knows the genetic sequences of scores of common respiratory viruses, the genetic sequences of Ebola and rabies and [name your horrific diseases of choice], and that can instruct people in the best way to insert one sequence into another.

It doesn't have to be perfect: one highly contagious virus with a double-digit mortality rate out of a few hundred attempts is more than good enough. Allah will be proud, or maybe some random Japanese death cult, or whichever nutjob is running North Korea.

I think those other dangers are real, but this one seems to me like the sweet spot on the plausibility/nightmare graph.