I still think there's a good chance that evolution has figured out some way to leverage quantum computation, probably in a very different way from the way we're trying to do it with ultra-cold low noise quantum digital circuits. If this is the case it's going to be some kind of high temperature noisy analog stochastic way of harnessing QC. The phrase "stochastic analog quantum computer" comes to mind.
It's how little energy the brain uses, especially for learning. The brain seems to be hundreds of thousands to millions of times more energy efficient than any kind of current AI on a classical computer, not to mention still beating it in terms of performance and versatility. Transistors do not use millions of times more energy than synapses, and processor feature sizes are not millions of times larger. Something else is going on.
Either the brain is leveraging QC or our AI training algorithms are just really really horrible compared to whatever is happening in biology. Maybe biology found learning methods that work thousands of times better than differential backpropagation.
I like the possibility of QC in brain. However, explaining why brain is much more efficient that computers does not need QC. Computers and Brain evolved in two completely different ways. For the brain, simple cognitive functions emerge first, supporting more complex life behaviours, starting with very simple multi celular life forms. Logical reasoning emerges much later, and is pretty expensive. Then we made computers to do logical computation and they are incredibly efficient at it: a modern low power processor is much more efficient than human brain in this kind of workload, by orders of magnitude.
Now we are trying to implement what the mind is naturally good at with systems designed to do logic well. This is the main reason it's so inefficient. Emulation is costly. It is costly when brain does logic, and is costly when computers do AI.
In theory, we should be able to build computing devices designed for AI workloads, and they can be as efficient as brain or even much better.
> The [human] brain seems to be hundreds of thousands to millions of times more energy efficient than any kind of current AI
I don't know about that... I've consumed quite a few calories in my lifetime directly, plus there is all the energy needed for me to live in a modern civilization and make the source material available to me for learning (schools, libraries, internet) and I still only have a minuscule fraction of the information in my head that a modern LLM does after a few months of training.
Translated into KWh, I've used very roughly 50,000 KWh just in terms of food calories... but a modern human uses between 20x and 200x as much energy in supporting infrastructure than the food calories they consume, so we're at about 1 to 10 GWh, which according to GPT5 is in the ballpark for what it took to train GPT3 or GPT4... GPT5 itself needing about 25x to 30x as much energy to train... certainly not 100s of thousands to millions of times as much. And again, these LLMs have a lot more information encoded into them available for nearly instant response than even the smartest human does, so we're not really comparing apples with apples here.
In short, while I wouldn't rule out that the brain uses quantum effects somehow, I don't think there's any spectacular energy-efficiency there to bolster that argument.
> plus there is all the energy needed for me to live in a modern civilization and make the source material available to me for learning (schools, libraries, internet)
To be fair, this is true of LLMs too, and arguably more true for them than it is for humans. LLMs would've been pretty much impossible to achieve w/o massive amounts of digitized human-written text (though now ofc they could be bootstrapped with synthetic data).
> but a modern human uses between 20x and 200x as much energy in supporting infrastructure than the food calories they consume, so we're at about 1 to 10 GWh, which according to GPT5 is in the ballpark for what it took to train GPT3 or GPT4
But if we're including all the energy for supporting infrastructure for humans, shouldn't we also include it for GPT? Mining metals, constructing the chips, etc.? Also, the "modern" is carrying a lot of the weight here. Pre-modern humans were still pretty smart and presumably nearly as efficient in their learning, despite using much less energy.
That’s why I said it would look nothing like the quantum-digital style of QC we are aiming at. It would be some analog stochastic way of leveraging quantum processes to accelerate information processing, possibly indirectly through their effects.
The brain and all biology is analog not digital. It’s really nothing like computers or discrete electronic circuits.
That’s what quantum computers are. I’m not sure what you mean by “quantum-digital style of QC” since the working quantum computers we have are very much in line with your analog stochastic information processing. And these machines do require very careful laboratory conditions to work.
FYI Penrose is a cautionary tale in the physicist community. He is/was once a competent academic, but his quantum consciousness ideas are viewed similarly to tinfoil hat conspiracy theories. It is technobabble word salad; quantum woo driven more by a personal objection to the implications of Newtonian determinism to the philosophy of the mind, not reason.
We understand quantum interactions more than sufficiently enough to know that the thread of hope he clings to, the soul of the gaps via quantum woo, is not in any way plausible. It is comparable to a perpetual motion machine.
You’re not providing me enough evidence other than saying it was just “woo” science. This is how a lot of new science is rejected.
If you can provide me with scientist objecting to the claim or writing negative papers about his theories I will gladly absorb them.
At best scientists are divided on his opinions, but this is far from calling them woo woo. I mean, bad scientist will call that but good ones will have honest disagreements and discussions.
> We understand quantum interactions more than sufficiently enough
Yes, we do understand quantum mechanics very, very well. It is a profoundly reliable theory, and fully describes all phenomena that could plausibly have causal effects on biochemistry. You have to get to black holes, galaxy formation, or femtokelvin above absolute zero to encounter regions where the standard model breaks down / new physics becomes possible.
Regarding critical evaluations of Penrose, this is the first that pops up: https://physicsworld.com/a/quantum-theory-of-consciousness-p... Like most published accounts, it is respectful towards Penrose and less inflammatory than what I wrote, at least on the surface. I'd draw your attention to this bit towards the end though:
> Still, they say, the overall requirements seem daunting – the brain needing to maintain a mass of 10−16 kg in a coherent state for 25 ms over a length scale of about 10 nm. “This vastly exceeds any of the coherent superposition states achieved with state-of-the-art optomechanics or macromolecular interference experiments,” they note.
This is a devastating statement hidden in technical terminology. Basically he's saying: "Even with the most sophisticated physics laboratories, under ideal conditions & with highly sensitive instrumentation, we're unable to achieve the superpositions that Roger Penrose is claiming is going on in the absolutely hostile thermal bath of the brain."
The brain 'stores' data without using power. Under classical synapse structure, it modifies the butons to modulate the charges and neurotransmitters passing and being received. This is memristance.
It's very low energy to do this and it keeps for decades (probably). It's not a quantum effect.
Be aware though, this is a 'classical' synapse understanding. The neurons are doing all kinds of other things too, they are alive after all. And the glia, the glia and astrocytes affect memory too, but we're still trying to understand how.
Look, don't jump to quantum stuff with the brain.
It's just really hard to get data, low sample sizes, and desperate need of grant funding.
The brain too learns from scratch. From birth through death, it's acquiring information, integrating that information, and using it. LLMs do this in a shorter time period.
It's how little energy the brain uses, especially for learning. The brain seems to be hundreds of thousands to millions of times more energy efficient than any kind of current AI on a classical computer, not to mention still beating it in terms of performance and versatility. Transistors do not use millions of times more energy than synapses, and processor feature sizes are not millions of times larger. Something else is going on.
Either the brain is leveraging QC or our AI training algorithms are just really really horrible compared to whatever is happening in biology. Maybe biology found learning methods that work thousands of times better than differential backpropagation.