> “The most important thing I’d want New York Times readers to understand is this,” Scott Aaronson, a quantum computing expert at the University of Texas in Austin, wrote in an email. “If this experiment has brought a wormhole into actual physical existence, then a strong case could be made that you, too, bring a wormhole into actual physical existence every time you sketch one with pen and paper.”
I love it. This "wormhole in a computer" business is a very long leap beyond the kind of is-it-or-isn't-it of, e.g., the "Chinese room".
Anyway, I didn't remember what that "Chinese room" business is, so I had to google it, and funnily, while I vaguely remember seeing it before and probably not having much of an opinion (as I didn't remember it), this time I got immediately annoyed by how such obviously stupid logical fallacy can deserve a Wikipedia page of its own and be a famous "thought experiment".
I looked it up because of what you said. It’s a pretty substantial article. What part of it is a fallacy? The part about computers obviously not having intelligence?
Searle is proposing what he supposes to be a proof of a computer being unable to have mind in the same sense humans have mind, and thus human intelligence being something more than a "mere computation". While in fact his argument proves absolutely nothing, and I'm honestly surprised that anyone buys into this bullshit. Actually, I think that the question itself is meaningless, but it's besides the point what I think, I'll just try to explain where's the fallacy. And since his argument is "reductio ad absurdum" it's easiest to counter by assuming the opposite, and showing that there's nothing absurd about it, i.e. there's no self-contradiction here. Just keep in mind that I'm not trying to prove the opposite, I'm only assuming it for the sake of argument.
So, the opposite assumption is that there's nothing more to intelligence than a "mere computation", and all that stuff like inner dialog and "a sense of self" are just side-effects of it. That way, there's no difference between "strong AI" and "weak AI" in his terminology, no difference between "understanding" and "simulation of understanding". Then, of course, a digital computer has a mind and understands Chinese, and also the Chinese room has a mind and understands Chinese. But Searle doesn't understand Chinese and this is supposed to be a contradiction.
But there's no contradiction here, because Searle is not the Chinese room. He isn't the mind, in this model he is to Chinese room what electricity is to a digital computer, or what neurotransmitters are to human brain. So, yeah, electricity doesn't understand Chinese, serotonin doesn't understand Chinese, single human neuron doesn't understand Chinese, but Turing-test-passing computer, Chinese brain and Chinese room all do. Searle is just a detail in the Chinese room — tiny and insignificant compared to a book, containing a complete description of a program that speaks Chinese and is able to form coherent (and persistent!) opinions on every possible topic in the world.
So, the fallacy is equating Searle's role in the hypothetical Chinese room to the whole digital computer that speaks Chinese and passes Turing test with breeze.
The Chinese room confusion shows how effective our deepest biases are at directing our rational thought into whatever level of nonsense you can put up with.
A similar case I noticed is the resistance against the Everett interpretation of the quantum measurement problem (i.e. the wave function collapse). For some people, it's so hard to accept the consequences of your own consciousness being an emergent phenomenon of matter that interacts with other matter and thus it partakes in the very fundamental effects of such interactions, including entanglement.
I think you've understood the idea perfectly, including all the ways in which Searle's argument is rubbish.
However, I come to a different conclusion. I think that all this is exactly why the Chinese room concept is interesting — because it is so obvious, but somehow induces a huge resistance in so many people. I.e. the thought experiment itself isn't all that interesting — people's reactions to it are.
The reason why Chinese room is a fallacy is because it's begging the question. It creates an elaborate metaphor for a computer only to exclaim that of course it's not intelligent, because that's obvious. If you don't start with a premise that "of course it's not intelligent", it's not at all obvious even for the analogy.
Searle's Chinese room is one of those things it's obvious most people talk about without ever having bothered to actually read. Searle does not say computers will never think. Early in the paper he even says - obviously machines can think because we are precisely examples of such machines. His point is that maybe it takes a certain kind of processing to yield what we call consciousness, and that may be a subset of the processes that yield outwardly similar behaviour.
I have read through the paper two or three times, as well as several other things Searle and others have written about it, and I have it open in front of me now.
In it, Searle argues that no computer program will be able to satisfy the goal he calls 'strong AI', which he defines thus: "according to strong AI, the computer is not merely a tool in the study of the mind; rather, the appropriately programmed computer really is a mind, in the sense that computers given the right programs can be literally said to understand and have other cognitive states."
As far as I recall (and I am not going to read the whole thing again just to check) he does not precisely define 'computer' or 'computer program', but it is clear that his meaning subsumes ordinary usage: a Turing-equivalent digital device, together with any program that can run on it. It is also clear that, given his underlying position being that syntactical manipulation cannot give rise to semantics, he would include, in his definition, actual universal Turing machines, and any other device, digital or not, that, in his view, can only perform syntactical operations.
Therefore, Searle is saying that computers will never think, except in a most pedantic reading of 'computers'. He is, indeed, a materialist, and regards the brain to be a mind-instantiating machine of sorts, but, like Penrose, he thinks its consciousness must depend on something beyond what we have discovered so far.
Personally, I think the argument begs the question in this sense: it is predicated on the assumption that the only thing in the room that could understand anything would be the human operator, but that, in turn, is predicated on the assumption that nothing else about the room could do so. That he is doing this is made very clear in his attempts to rebuff the 'systems reply' and the 'simulator reply'.
I'm not going to pretend to be an expert, but based on looking this up it seems like the systems reply is fairly convincing a response - Searle's argument seems to be, as others have noted in this thread, kind of at the level of saying the electricity and chemicals don't understand Chinese and then extrapolating it to 'no understanding is present'. How does that not apply to us? Otherwise if that's all he's saying it doesn't seem like a particularly interesting argument, because I'd agree that a chemical doesn't understand Chinese since our understanding isn't generated at the singular chemical or even singular neuron level.
What's your take on this (very rapidly acquired) impression?
Imagine a brain, it's made of two parts. Say, chemical impulses, and a physical substrate. Now the chemicals don't speak Chinese. And the physical substrate doesn't speak Chinese.
There's no "part about computers obviously not having intelligence"
The most grating part of the article is involving political names into the explanation of the argument, and that the writings by Liebniz and Dneprov seemingly would offer far more interesting material to base the topic on.
Wikipedia can be terrible source of misaligned atheist theology sometimes.
Someone recording a concert 50 years ago, that concert now being freely available on Archive, and me blasting it on my home stereo, via Bluetooth, from my wirelessly charged pocket computer with wifi is a wormhole. And also a miracle.
The two senior physicists behind this, Joe Lykken and Maria Spiropulu, have histories that go way back of successfully promoting to the press nonsense about exotic space-time structures appearing in experiments that have nothing to do with them.
and
Update: Andreas Karch on Twitter I think has an accurate characterization of this “mostly a publicity stunt”:
Experimentally it’s of course cool they can do SYK – as a demonstration they have control over their device. They can couple 9 qbits in a pre-specified way. But I guess we knew they could do this before. Going after SYK in particular, in my mind, is mostly a publicity stunt.
This is a very cool and important result. Here's a rough analogy to what's happening. You take two drums, and you hit one of them. Then you wait for a while. After waiting, you hook up a special kind of wire between the drums, and then quickly unhook it. Then, you wait the same amount of time again. And bizarrely, you hear the original sound coming from the other drum!
How did this happen? It seems like no wire would be capable of moving the excitation from one to the other in a clean way. But if the system is set up correctly, it's because your "drums" obey identical mathematics to a pretty interesting gravity setup. Each drum individually represents a region of space, and the two spaces are initially unconnected. The wire is a way of establishing a traversable wormhole between them. These wormholes are very unstable, so to actually send something through it is a delicate affair. But with the right kind of entanglement, you can send the excitation through, to emerge on the other side after a propagation delay. (Nothing is happening "faster than the speed of light", though, even in the gravity setup. It takes time to pass through the wormhole. It's tricky to show there's no problems with causality, but the "wire" is already messing with the ordinary story about causality, so it does all work out.)
This is a lovely example of how string theory, via AdS/CFT, can produce fascinating insights into the nature of spacetime. It's an extremely nontrivial bit of stringy math that leads to this connection between quantum mechanics and gravity. Of course, if one's prior is "string theory can never be useful for anything", it makes sense to downplay the importance of the result. I think that's a shame--AdS/CFT is an extraordinarily fascinating and powerful tool.
(I didn't know this result was coming out today, but I did talk excitedly about the possibility of this experiment earlier this week on this very website! https://news.ycombinator.com/item?id=33735328 )
Since this comment was more controversial than I expected, let me add a few notes.
1. Everything done on the quantum computer here can be (and was) also simulated on a classical computer. The most interesting part of the story here is the algorithm, which is derived from a string theory argument.
2. Even so, since entanglement plays a key role, a quantum computer is way more natural to implement this on than a classical computer.
3. There are some profound philosophical consequences of AdS/CFT. A lot of the discussion about simulation vs reality here would be better served by taking some time to deeply understand the topic in detail, because it is extremely counterintuitive and interesting. The argument that “everything done on a computer is by definition a simulation so it is not interesting” is far from the end of the story.
> I just saw that the New York Times also has a big story about this: Physicists Create ‘the Smallest, Crummiest Wormhole You Can Imagine’. At least this article has some sensible skeptical quotes, including:
>> “The most important thing I’d want New York Times readers to understand is this,” Scott Aaronson, a quantum computing expert at the University of Texas in Austin, wrote in an email. “If this experiment has brought a wormhole into actual physical existence, then a strong case could be made that you, too, bring a wormhole into actual physical existence every time you sketch one with pen and paper.”
"A better summary: Physicists created an entangled system of qubits in a quantum computer in a way that is well understood in conventional quantum mechanics, and some physicists have speculated that entangled particles are connected by wormholes."
I'm sad to say I used to really like Quanta magazine pieces but I've become adept at knowing immediately from a Hacker News discussion title (the direct links, not this one) that it's about a Quanta piece, and it's going to be a letdown when you drill in.
I'm not sure if they've become more clickbaity or less discriminating, or if it's just me moving on.
All accounts seem pretty sneering and dismissive. Quantum wormholes,
whether they exist of not are quite beyond me. But the gap in
space-time that does interest me is the one between theoretical
physicists and application engineers. The former seem to crave the
light of publicity while the latter inhabit dark university basements.
The theoreticians may as well have a publicist, broadcasting weekly
"Oh hi. We're still here. Haven't discovered anything actually useful
yet. But please keep sending grant money".... And then boom! Suddenly
someone invents a transistor, or laser from the stuff everyone was
mocking as unhinged science fiction.
I think there are very, very few people who can make the connection
between these worlds, and even less of them are science journalists.
I love it. This "wormhole in a computer" business is a very long leap beyond the kind of is-it-or-isn't-it of, e.g., the "Chinese room".