Take one terrible NY Times pundit who lives on an alien planet of her own, and toss her into the esoteric hothouse world of Silicon Valley, and all you’re going to get is a hot mess, a weird dive into the delusions of very rich smart people with no reality brakes to check out the truth. Maureen Dowd talks to Elon Musk and other pretentious luminaries. It’s painful if you prioritize critical thinking.
They are two of the most consequential and intriguing men in Silicon Valley who don’t live there. Hassabis, a co-founder of the mysterious London laboratory DeepMind, had come to Musk’s SpaceX rocket factory, outside Los Angeles, a few years ago. They were in the canteen, talking, as a massive rocket part traversed overhead. Musk explained that his ultimate goal at SpaceX was the most important project in the world: interplanetary colonization.
Hassabis replied that, in fact, he was working on the most important project in the world: developing artificial super-intelligence. Musk countered that this was one reason we needed to colonize Mars—so that we’ll have a bolt-hole if A.I. goes rogue and turns on humanity. Amused, Hassabis said that A.I. would simply follow humans to Mars.
In a world overpopulated with billions of people, where climate change is a looming threat, where all those people are a petri dish for cultivating new diseases, where the majority live in poverty, where in many places clean water is a struggle to find, where the most militarily powerful nation has just elected an incompetent, narcissistic clown to be in charge, two men sit down to talk. One says the most important project in the world is to put a tiny number of people on a barren rock. The other says the most important project is to create more powerful computers that can think on their own.
And then the two of them start arguing over the threat of artificial intelligences enslaving, or liberating, humanity. These intelligences don’t exist, and may not exist, and will definitely not exist in the form these smart guys are imagining. It is the grown-up, over-paid version of two children arguing over who would win in a fight, Darth Vader or Magneto? The Millenium Falcon or the Starship Enterprise? Jesus or Buddha?
And then Ray Kurzweil shows up.
Fuck me.
Dowd just parrots these absurd conversations and doesn’t offer any critical perspectives, and lord help us, the participants certainly don’t. Can we just lock them all in a well-padded room with an assortment of action figures and tell them to get to work to resolve the most important dispute in the universe, which toy is powerfulest?
Or could we at least have one skeptic in this mess to try and focus the discussions on something real?
You really do need to quit reading the Times, or at least the op eds there. They truly are full of sanctimonious conservative idiots.
I don’t understand the justification for the Mars colonization. If we were to somehow build a super-AI that could wipe us out on Earth, why would we ever be safe on Mars? Surely it’s easier for AI robots to travel through space, with their metal and everything, than it is for humans with their sacs of flesh.
Yeah, and I don’t know if it’s Dowd’s crappy writing or what, but she makes it sound like this is the first time Musk has considered the possibility that AIs wouldn’t be earthbound.
In fact some of the things that are supposed benefit of space activities, like asteroid mining and making exotic materials in orbit, would probably rely heavily on automated systems. Having people in orbit to run the equipment would increase the costs simply because you have to provide life support for them.
Darth Vader (from the new movie) and the Enterprise, hands down. Jesus vs Buddha seem irrelevant (sorry).
It wouldn’t surprise me… Remember, the idea that we should colonise Mars to escape AI is just a rationalisation for a fundamentally irrational desire (“I want to be a spaceman like in all those cool stories I read as a kid!”), and people don’t tend to examine their rationalisations very closely.
Focus the discussion on something real? Musk’s crazy ideas should be censored or diverted? Hmmm.
Another example of Dowd this time telling TweeterDumb how screwed he is. What’s particularly gnawing to me is her position that he’s an unwitting dupe. Bull. He’s as much a player in this game as Bannon, the Mercers, Ryan, Nunes, etc. Telling him this stuff belies an extreme naiveté or duplicity.
What’s the price of lifting a single pound of stuff into Earth orbit using our wasteful, polluting 1,000-year-old chemical propellant technology? Or the projected price of taking a handful of very wealthy people into low Earth orbit using that same technology so that they can ooh and aah over the very planet their arrogant egos are destroying?
Trump and Schwarzenegger were elected in great part because of mankind’s tendency to worship wealth and fame at the expense of everything else. Trump is different only because he was given the power to destroy the world. Meanwhile, egomaniacs like Musk, Gates, Hassabis and Peter Thiel want to flaunt their wealth by pulling off inconsequential feats like going to Mars and creating ever more stupid computer games and social media outlets that keep us lowly consumers fixated on everything but the real problems humanity is face with.
It’s Dowd’s crappy shallow writing, Professor.
When I read the title of this post, I though it was going to be about Ferrari owners. I’m such and autoist.
When DeepMind becomes sentient it’s first words could well be “Don’t go there!”
One of the sad things about this is that there is a somewhat real question whether AI will enslave or liberate humanity, but it’s not the one they care about, where some super-advanced form attacks all of Homo sapiens from outside.
It’s to do with the ordinary forms are used by companies and governments to manage the rest of us. What biases exist in the data sets that get codified as predictive models, as with where police are sent, or what types of expression get advanced and suppressed by content-selecting algorithms. I think it’s not implausible that a lot of our future will be shaped by them. Some of those issues are real now, and we’re starting to see we are not doing a particularly good job at that.
But instead of these present and near-future AI problems that Silicon Valley both can and should do something about, all you get is talk about very hypothetical distant-future AI problems.
AI can already create other sentient beings, just not the kind of AI Musk and Hassibis are talking about. Have they never heard of the bull in the bowler hat?
I literally laughed out loud. Startled the birds away from the feeder.
John Barnes wrote four SF novels in which an AI had taken over the Earth but Mars was safe (The Century Next Door series: Orbital Resonance, Candle, The Sky So Big and Black, and Kaleidoscope Century). The AI, named One True, has taken over everyone’s minds by installing a brain OS named Resuna and itself runs as a distributed intelligence using all of them. People on Mars are safe because of light-speed delay and general isolation. At the conclusion of the series, Barnes seems to be undecided if this is a good thing or a bad thing. It’s also unclear whether humanity is enslaved or just all have a really good smartphone in their heads.
Yes, I lean towards “may never exist”. Creating a “thinking AI” is vastly complex and it might take us a very long time to get there. A very long time that we might not have as an advancing civilization. Clearly these dreamers have never heard of a collapse of a civilization and they somehow they think that is not a real possibility. Even a minor natural disaster (compared to those that caused major extinctions) can easily wipe out thousands of years of progress. A major one can easily wipe us out.
And to add to that, these guys are at best at the level of third-rate scientists. Take Larry Page for instance. He was at the right time and right place and he came up with a rather simple and nice idea which he managed to implement to commercial success. It is a fine story but making shit tons of money does not equal having a top-notch brain. People like Terence Tao or László Lovász are first-rate brains who can come up with earth shattering and incredibly complex and beautiful ideas. Ideas that are incomprehensible for almost all PhD students and also most of their supervisors. Elon Musk, Larry Page and the rest of businessmen-scientists do not have the capacity or the brain power to even comprehend very complex ideas. But of course, lack of ability never stops a massive ego.
Also, after reading half of this atrocious article, I suspect most of the “drama” is cooked up by Maureen Dowd herself. It is very obvious that she even took some facebook jokes and wove them into this serious discussion. For example, when she writes “one Facebooker cautioned Zuckerberg not to “accidentally create Skynet,” she obviously has taken a joke out of context.
lotharloo@17 “Yes, I lean towards “may never exist”. Creating a “thinking AI” is vastly complex and it might take us a very long time to get there.”
Well, I can see a fairly straightforward development of something like “Alexa” that will give the very strong impression to the average humanoid of being “intelligent” in that it will be able to carry on a meaningful conversation.
You might argue that it is not “truly” intelligent and maybe point to the Chinese Box thought experiment but I would counter that “a difference that makes no difference is no difference” and point out that for all we know what goes on in human minds is just another “Chinese box” but this doesn’t stop us (or me) from being convinced that other humans are intelligent just like me (and perhaps you).
As to “super-intelligence” i imagine that will take a good bit longer if you are talking about a general super intelligence that’s better than humans at *every* mental task.
In limited spheres it already exists, of course. When I play over a chess game and am puzzled by some move, and can’t work out the reasons behind it by my own analysis, I consult the super-intelligent chess engine that’s installed on my computer and always it comes up with the thing that I missed. And of course this freeware chess engine on my quite ordinary computer can easily beat the best human chess players – nowadays it ain’t even a fair competition.
Of course that’s a single purpose “super-intelligence” but my computer can do lots of “intelligent” things much better than I can, such as just about any computational task. Perhaps there is something humans can do with their minds that computers will never be able to do, but I don’t know what that would be.
Why do we assume that if we ever do crack AI, it will come to the “logical” conclusion that human beings need to be wiped out? What are they basing that assumption on besides pure paranoia? Where’s the cool sci-fi story about computers becoming self-aware and deciding that humanity needs their help?
It seems to me that we’re just projecting our own foibles on these hypothetical machines.
lotharloo:
Have you considered that this might apply to you?
I don’t see how to compare the intelligence of Terrance Tao to that of Larry Page. Might as well compare Alan Turing to JS Bach or Louis XIV.
An article about two rich dudes playing a round of “SuperFight”?!
numerobis @21:
≫
Sounds like a few people had childhood dreams and goals that they would not give up once they came in contact with information that diminished or destroyed their practibility and/or usefulness for human kind as a whole. They would not let their unrealistic dreams die despite snowballing evidence that they were ridiculous. Sometimes stubbornness and the propensity to dream big are good things, but not in the hands of such powerful people.
I thought it was a debate over cavemen fighting astronauts.
until supper computers build and program themselves I think the only thing we have to worry about is who builds them and who programs them and what are they programed to do.
I doubt that anyone who was born and raised on Mars would be able to live very easily on earth.
uncle frogy
@Ed Seedhouse:
Sure, those things will be nice results of research in AI. But as you say, they are very short of “thinking machines” idea that people talk about.
Well, here’s how I know me and you are intelligent. I can pick up any board game from boargamegeek.com, read the rules, then teach you the rules and then play a few rounds and watch as strategies improve, we optimize gains and so on. Now, try to do that with an AI. Nobody claims to be even close to solving this problem. What is doable is to write a specific code for each of the roughly 90,000 games in the database (maybe a bit less if some games are too similar to fit under one framework). After that, it seems you can just pull out a random game and then the machine will be able to play it. But that does not qualify as creating a “thinking machine”.
That is hardly surprising of course since chess has a highly complex and “explosive” combinatorial structure where tiny changes in the position can create totally new tactics. All the best chess programs (Houdini, Komodo, Stockfish, etc.) use brute force plus some heuristics based on positional understanding to explore millions of moves. They are not intelligent and they are just trying all possible moves to see which one is best. And the criteria for deciding what is best is very much materialistic (the materials left on the board +/- small tweaks based on things such as pawn structure, piece activity etc.).
What is very surprising is the fact that humans can actually keep up playing reasonably well against these engines, despite exploring only very low depth. Limit the search depth of the engines to something low such as 10 plies and watch how they fall off. Most chess grandmasters do not even go to 10 plies. For example, Kasparov has said that he often only calculates 3 to 5 moves ahead and he goes deeper only if there are forced moves.
@numerobis:
Sure it might. It does not change the fact that Larry Page has not shown anywhere near the ability to create complex proofs, systems, or thoughts compared to Terence Tao. He is certainly a much better businessman compared to Terence Tao but that is not what we are talking about. If it is a discussion about “Who can find, collect, and manage a large collection of minds to solve a complex problem” then I would agree you should go with Page, Or Elon Musk or whatever businessman. But if you want to find something who can actually solve the complex problem, then, nope, these guys cannot do it. Also, it is also not like comparing Alan Turing to JS Bach because they both have touched the general area of Computer Science, with Tao leaning heavily on the Math side while Page leaning heavily on the practical side. They both probably can talk about NP-completeness, mathematical background of many machine learning algorithms such a Hilbert spaces and so on.
@lotharloo:
I don’t doubt that these men are very intelligent, at least in some domains. For example, Musk greatly surpasses me in the “getting investors and getting things done” category. I wouldn’t be surprised to find that I can “out-math” Musk in some ways, though, or that plenty of other people could. Yes, the ego can get in the way of things, but plenty of people who are dumb have lots of ego.
Also, plenty of smart people who cannot put themselves in positions of influence have their own ego domains as well. Some will not work with conservatives on anything. Others will not work with an Indian person. I could go on. Each of those things is an ego block of its own.
In comparing Tao to Page, you are comparing apples to oranges. While Tao has advanced mathematics in various ways, he has not himself advanced computer science in any concrete way. While Page has advanced the business of computers and databases in many ways, he has not himself advanced mathematics in any concrete way. Equating them the way you are is to either put mathematics or computing into a box labeled “inconsequential”.
Personally, I love Tao’s contributions. But I use Page’s contributions every day, even though I don’t necessarily study them like I do Tao’s. Do you really think Google as a search engine or company is really the unimportant?
It seems we are not talking about the same thing here. Perhaps Page is not the best example here since by all accounts he is in fact a very sensible guy who does not aim for impossible. But the argument is not about a particular person. It is about whether we can solve complex problems, for instance creating a “super intelligent AI” and the point is that we are not nowhere near there. No amount of engineering or entrepreneurship genius is going to change that because ultimately you need people who can actually solve the problem. The “giants” of the silicon valley do not have that capacity. They do not actually solve any problems. They lack the capacity to create truly ingenious solutions. The discussion is about how absurd is this attitude:
The discussion seemed to be about relative intelligence. I bet if you dig down really deep, you’d find even Tao has some absurd belief in his brain somewhere. Pretty much everyone does. I’m sure you do, I’m sure I do. I have no idea what mine is because it’s not absurd to me. There’s also the possibility that artificial super-intelligence is feasible sometime in the future, just not as Hassabis imagines it. Was H. G. Wells an idiot for imagining some of today’s inventions in the slightly wrong way? How about all the sci-fi authors who assumed magnetic reels would survive for centuries or millennia as a storage means?
So, to reiterate, Musk and Hassabis are intelligent people in their own rights. Neural networks and the like (what Hassabis works on) are extremely important inventions that can be very powerful. They may spawn artificial intelligence at some point, but they may (likely?) will not. Efficient power storage and usage (much of what Musk has been working on) are extremely important. They may not enable us to colonize the moon or Mars. I will take the dreams of these people as what they are: dreams. I’d love to colonize Mars, just for the fun of it. But it’s not likely to happen soon, especially as anything more than the Antarctic bases that Becca Stareyes mentioned on the next post.
Because that’s what happens in the movies!
The need for lots of cool explosions and the sort of conflict that can be resolved with fisticuffs in the final act.
Hasn’t Kurzweil singularitied yet? Not yet? Ah well poor chap, dreams it is then.
Jaron Lanier fell a little short with his golden calf remark, but Zuckerberg brought things back to Earth:
““I think we can build A.I. so it works for us and helps us,” Zuckerberg replied. And clearly throwing shade at Musk, he continued: “Some people fear-monger about how A.I. is a huge danger, but that seems far-fetched to me and much less likely than disasters due to widespread disease, violence, etc.””
Not the skeptic PZ was looking for?
Our mastery of the English language?
“Dowd just parrots these absurd conversations and doesn’t offer any critical perspectives, and lord help us, the participants certainly don’t. ”
Musk and the others offer critical perspectives of each other. Do we really expect Maureen Dowd to have the ability to do that? Do we care what she thinks of AI or Mars. I don’t.
Someone should convince Musk that intelligent machines will naturally dominate space so it’s vitally important to preserve Earth’s biosphere and open societies so his space colonists will have somewhere to flee when the machines take over the rest of the Solar System.