The Case Against Quantum Computing

Ya’akov

Joined Jan 27, 2019
9,079
Roger Penrose has stated that a digital computer algorithm can never generate true consciousness, and suggested that the brain uses quantum processes in the brain cells microtubules to provide our consciousness, and the leaps in knowledge that allows.
The Emperor's New Mind was speculation and further research has made the theory extremely farfetched. A lot of other things, in fact everything that leads up to the speculation, is well worth reading. It is a very cogent exposition of many useful ideas.

Penrose just didn't have the benefit of the later quantum research to work with.

[EDIT: typo repair]
 
Last edited:

bogosort

Joined Sep 24, 2011
696
My personal opinion, is that true AI will only be accomplished when QC finally goes mainstream. I think that calling current complex processing algorithms "artificial intelligence" is doing a disservice to the concept itself.
I agree with you that there is very little (if any) intelligence in AI, but I disagree that accessibility to QC would somehow change that. Perhaps Penrose is correct that consciousness is a quantum phenomenon, but even if that were true, it seems orthogonal to QC.

The single, special ability -- if indeed it is special -- of QC is to leverage interference between the superpositions of n qubits to solve a problem that would otherwise require 2^n classical bits. In other words, the great promise of QC is to provide exponential speed-up to the very specific class of problems whose solutions can be found through quantum interference. Personally, this doesn't seem relevant to the challenge of creating artificial intelligence.

I think this becomes clearer with a slightly different framing: suppose we were able to create intelligent machines with classical computing, but they were exceedingly slow thinkers. Then, QC might be able to help.

It's important to remember that quantum computing does not promise magical abilities; everything that a QC can do, a classical computer can also do (just slower).
 

cmartinez

Joined Jan 17, 2007
8,220
everything that a QC can do, a classical computer can also do (just slower).
I'm not sure that statement is correct ... then again, I lack the background and info to objectively contradict you on that ... To me, it's a fascinating topic for sure.

Here's my take: I believe that the human brain is fundamentally a quantum parallel computer (don't know if the word "parallel" is redundant, though) and somehow can make multiple simultaneous (and possibly unrelated) events coalesce within its "working algorithm" and hence produce consciousness.
 

crutschow

Joined Mar 14, 2008
34,285
But that suggestion so far amounts to mere speculation (albeit from a very respectable figure), right?
Yes, but it is seems better than anything else we have to explain consciousness.

Gödel’s theorem shows that certain claims in mathematics are true but cannot be proven. “This, to me, was an absolutely stunning revelation,” Penrose said. “It told me that whatever is going on in our understanding is not computational.”

So if the brain is not strictly computational, then a computer can never achieve true consciousness or be self aware, and the possibility that quantum effects in the brain cells contribute to consciousness is an interesting theory (and there are no other viable theories I know of).
 

bogosort

Joined Sep 24, 2011
696
I'm not sure that statement is correct ... then again, I lack the background and info to objectively contradict you on that ... To me, it's a fascinating topic for sure.
I suspect it is a common misconception that QCs can somehow produce computations that a regular computer can't. But if you think about it carefully, I think you'll realize the impossibility of such a feat.

QCs work just like regular computers: they take an input signal and pass it through a set of gates to produce an output signal. Quantum gates are indeed different from classical gates, but we can nonetheless describe what do they do. And if we can describe what they do, then we can always write a classical computer program to execute that description. Hopefully that algorithmic aspect is clear, but we can hash it out further if you want.

Now imagine the alternative scenario: QCs use quantum gates whose behavior we somehow cannot describe. How would we even program such a machine? How would we know if it were behaving correctly? A gate or operation that cannot be described is useless for computation.

Here's my take: I believe that the human brain is fundamentally a quantum parallel computer (don't know if the word "parallel" is redundant, though) and somehow can make multiple simultaneous (and possibly unrelated) events coalesce within its "working algorithm" and hence produce consciousness.
Another common misconception is that QCs can somehow "check" every possible solution in parallel to give us the correct answer. That's not at all how they work, but notice that even if it were true, we can certainly build a network of parallel-processing classical computers to do the same thing (though perhaps slower).
 

bogosort

Joined Sep 24, 2011
696
Gödel’s theorem shows that certain claims in mathematics are true but cannot be proven. “This, to me, was an absolutely stunning revelation,” Penrose said. “It told me that whatever is going on in our understanding is not computational.”
Gödel’s incompleteness theorems are statements about formal theories and their languages. They do not in any way imply that the brain is not computational.
 

bogosort

Joined Sep 24, 2011
696
Recent research, has indicated that bird navigation involves quantum processes, so it's not above the realm of possibility that our brain does also.
Quite literally everything, including the chemistry in our brains and the transistors in our computers, relies on quantum processes. In the context of QC and AI, the relevant question is whether or not QC has some property missing in CC that "enables" GAI. Given that QCs are computational devices, Penrose would presumably think not.
 

Thread Starter

nsaspook

Joined Aug 27, 2009
13,086
'Real' intelligence runs on wet-ware at room temperature. There is a solution to general intelligence that doesn't require current 'pure' QC architecture. Saying something is quantum doesn't mean very much as the classical transistor junction is quantum as was said above.
 

Ya’akov

Joined Jan 27, 2019
9,079
Recent research, has indicated that bird navigation involves quantum processes, so it's not above the realm of possibility that our brain does also.
There is suggestion something at the quantum level is involved, but if it is, it couldn’t be the sort of process Penrose describes in The Emperor’s New Mind.
 

cmartinez

Joined Jan 17, 2007
8,220
Gödel’s theorem shows that certain claims in mathematics are true but cannot be proven. “This, to me, was an absolutely stunning revelation,” Penrose said. “It told me that whatever is going on in our understanding is not computational.”
Interesting ... but Gödel gave proof that some things cannot be proven. I wonder, is there a way to prove that "our understanding is not computational", or to prove that it cannot be proven?
 

Thread Starter

nsaspook

Joined Aug 27, 2009
13,086
"Since you all seem to be smarter than Penrose"
:rolleyes:
The problem with Penrose is not how smart he is. The things he says about QM consciousness are flatly inconsistent with QM as we understand it today. Everything we know about QC today says that quantum mechanical processes (quantum effects in proteins known as microtubules) in the brain would lose coherence long before the cell or nerve level. IMO he proposes no new ideas on how this would be possible or even probable with advances in science.
Penrose has an opinion about objective reality and intelligence, we all do. He is an original thinker.
https://royalsocietypublishing.org/doi/10.1098/rsta.1998.0254

https://arxiv.org/pdf/quant-ph/9907009.pdf
The Importance of Quantum Decoherence in Brain Processes
 
Last edited:

Ya’akov

Joined Jan 27, 2019
9,079
Since you all seem to be smarter than Penrose, I'll bow out.
It’s not a matter of being ”smarter”. Penorse wrote The Emperor‘s New Mind 33 years ago! There has been a lot of research since then, of course. Some of the basis for his (self described) speculation was shown to be untenable. This is not because of some defect in Penrose, it’s because science progresses.

We (collectively( know a heck of a lot more about quantum effects and the energy levels in play, etc. Since the writing of the book, many researchers have shown the impracticality of the mechanism Penrose suggests. This doesn’t mean that there isn’t something important happening at the quantum level, it just means there is currently no evidence that it is required.

The main reason to consider Penrose, even now, is the problem of timescale. While I am really not convinced that you couldn’t have an algorithmic analog to human intelligence, it is clearly not how the brain is operating at the level we can observe, It doesn’t consume enough power, nor have a high enough “clock rate” to be a Von Neumann machine.

Roger Penrose is an incredibly intelligent and clever person. I really don’t know what his current thinking on this is, (I have searched quickly and found nothing current from him) but I have to imagine he’s refined his ideas in light of the new informatiion.

This doesn’t mean , by the way, that he would have abandoned his idea that consciousness is in some way distinct from what neuroscience can observe. His focus on human experience is something for which I have great sympathy. The problem of things like qualia are often simply rejected by researchers and philosophers as “malformed questions” but they refuse to go away. The question of how consciousness can arise from the dirt we are made of is a vexed one. Ideas like panpsychism seem promising as an answer. I tend towards that in my own philosophy.
 

bogosort

Joined Sep 24, 2011
696
While I am really not convinced that you couldn’t have an algorithmic analog to human intelligence, it is clearly not how the brain is operating at the level we can observe, It doesn’t consume enough power, nor have a high enough “clock rate” to be a Von Neumann machine.
How are you determining sufficient power consumption? If we take the upper bound of computational power usage as the black hole threshold (Bekenstein bound), and the lower bound as the Landauer limit, then I think we can all agree that human brains are nowhere near either limit.

In terms of time, I've never heard of a lower bound or minimum clock speed condition. (I suppose the potential heat death of the universe is an effective lower bound on clock speed.) But given that human brains can indeed compute, it would seem that our clock speeds satisfy any possible minimum requirements.
 
Top