Software-only machines are qualitatively different than machines with paper trails, since they can be tampered with en masse by a small number of adversaries without needing physical access at voting time; and because such tampering may not leave any trace whatsoever.
You’re right. I didn’t read the paper very carefully, and was myopically focused on the emulating-a-real-brain AI strategy. As in, let’s slice up a real human brain, map the neurons and synapses, and then simulate them as faithfully as possible.
To do that you need a great deal of fidelity in your simulations of neurons, which are enormously complex. But there is an argument to be made that neuronal complexity is incidental to the brain’s overall “computational capacity”; that you could replace the neurons in a human brain with much simpler nodes and still end up with a functional intelligence, after sufficient rewiring.
I don’t think that claim is obvious, but it’s definitely possible. And if it’s true, you can have human-level intelligence for 10^19 ops, given suitable software.
So I apologize for my post. It was over the top and unfair.
All that said, I still disagree with Bostrom’s conclusions. I think he enormously understates the difficulty of creating intelligent software, if we’re not just copying an existing brain.
Let's look at the most important section of the paper. He estimates the processing power of the brain:
The human brain contains about 10^11 neurons. Each neuron has about 5 • 10^3 synapses, and signals are transmitted along these synapses at an average frequency of about 10^2 Hz. Each signal contains, say, 5 bits. This equals 10^17 ops. The true value cannot be much higher than this, but it might be much lower.
In other words, there are 5 * 10^14 synapses in the brain, and each synapse transmits up to 100 signals per second, and we can probably encode each signal with 5 bits. That's ~10^17 bits per second.
So, uh... does anybody else notice that that's not an estimate of processing power?
That's an estimate of the rate of information flow between neurons, across the whole brain.
The level of confused thinking here is off the charts. Does this guy not understand that in order to simulate the brain, you not only have to keep track of information flows between neurons, you also need to simulate the neurons themselves?
That's not merely a flaw in his argument. It indicates that he has no idea what he's talking about, at all.
Needless to say, this paper and its conclusions are complete nonsense.
Well, if it's possible to build a human level intelligence, it's probably possible to build an intelligence that's much like a very smart human except it runs 100x faster. And in that case, somebody with sufficient resources could build an ensemble of 1000 superfast intelligences.
That's a lower bound on the scariness of AI explosion, and it may already be enough to take over the world. Certainly it should be enough to take over the Internet circa 2015...
To my mind it seems pretty clear that if AI exists, then scary AI is not far off.
That said, I don't worry about this stuff too much, because I see AI as being much technically harder, and much less likely to materialize in our lifetimes, than articles like this suppose.
My understanding is there are two reasons you want inflation:
(1) Inflation means real interest rates can go negative, which means it's harder to get stuck at the zero lower bound (as we are today).
(2) Wages are sticky, i.e. workers fight very hard against having their wages reduced. When there's inflation you can effectively cut wages by giving raises below inflation. Without inflation it's very hard to cut wages, and the economy becomes brittle.
> My general impression is that more and stricter regulations in an industry is usually (though not always) a bad thing for consumers in that industry
In cable internet the average consumer has only one or two potential providers. On top of that, companies who try to break into the industry face significant political and regulatory challenges when they lay fiber.
I don't understand how you can think this kind of industry should be unregulated. There are few market forces keeping providers in line, and we see the results of that in the industry's attempts to extort internet companies with throttling.
No, booking a ride with the intention to cancel is definitely wrong. My point is that Uber employees aren't doing that frequently enough to affect Lyft at all, so this can't be a plan on Uber's part.
It sounds like some Uber employees are just being assholes for no reason. That's unfortunate, but it's no reason to malign Uber itself. Every company has some assholes.
Lyft says that Uber employees have canceled 5000 rides since last October. That's like $100k in lost revenue -- it's nothing. Definitely not some big plan by Uber to undermine Lyft.
Aggressively poaching drivers from Lyft seems all right to me. That's just healthy competition.
And this stuff about wasting Lyft's money by booking and canceling rides -- it's a bunch of nonsense. By Lyft's own estimate, Uber employees booked and canceled 5000 rides since last October. That's like $100k/yr in lost revenue. It's nothing. Certainly not a systematic attempt to undermine Lyft.
Am I missing something here?
Disclaimer: of the two services I only use Lyft. I've had bad experiences with Uber's customer service.
If the fixed income came from newly-printed money then yes, it would cause inflation. If it came from taxes then no, it would not cause inflation in and of itself.
In both cases, it would increase the purchasing power of the poor and reduce the purchasing power of the rich, because $5000 means more to somebody making $10000 than it does to somebody making $1000000.
Other industries are deeply unlike tech, so comparing salary trends would be meaningless.
A sibling comment suggests that we look at salaries in tech companies that didn't collude. But that doesn't work either, because (1) those salaries could have been indirectly affected by the collusion, and (2) other tech companies are different than Apple, Google, et al -- most importantly, they have less money!
If, in my nonfiction book, I write, "Under normal circumstances, water boils at 80 degrees celsius," then I am a liar, and there is nothing I can say after the fact that will excuse my behavior.
Malcolm Gladwell often writes lies in his nonfiction books. I have found myself personally deceived by some of his lies. For instance, on page 39 of Outliers, he writes:
The striking thing about Ericsson’s study is that he and his colleagues couldn’t find any “naturals”, musicians who floated effortlessly to the top while practicing a fraction of the time their peers did. Nor could they find any “grinds”, people who worked harder than everyone else, yet just didn’t have what it takes to break the top ranks.
The first time I read this passage, I thought it was one of the most extraordinary pieces of information I'd ever encountered. It meant that practice almost completely determines ability! I mostly believed that for more than 4 years. (And so did the rest of the world -- that's the core of the 10000 hours meme.)
It turns out that the passage is completely false. Ericsson's study did not actually say anything about individual violinists at all -- it only ever reported average statistics of groups of violinists.
Other studies of the relationship between total practice time and ability have found that different people may reach a given level of ability with enormously different practice totals. For instance a study of chess players found that the average amount of practice time needed to reach the Master level is ~11000 hours -- but the standard deviation of the distribution is more than 5000 hours. One person only took 3000 hours to reach Master level. Another took 23000 hours. [1]
Calling this kind of deception "being playful with ideas" is absurd.
Do you really not see how that’s a big deal?