I'm not commenting on the claims of the article, just the points in your comment.

I think it's pretty clear that an entire human brain is not required for the operation of consciousness. Certain people have lost massive portions of their brain and still maintained regular consciousness functioning. On top of that, the typical human brain only consumes about 20 watts of power to do its thing. So that's just a safe upper bound for the power required.

Over the last decade we've seen the rise of ML systems that have replicated or surpassed capabilities long thought to be the exclusive domain of the human brain. Think facial recognition, AlphaGo or the very recent DALL-E 2.

This had led me to the personal belief that we (in aggregate) likely already have the computing power to achieve not only artificial consciousness, but also AGI and beyond. We simply haven't figured out the correct model connectivity and parameters.

I think that you need more than just computing power, you need high levels of interconnection with low latency.

So while we undoubtedly have sufficient aggregate computing power, I don't think it is structured in a way that would allow it to emulate human cognition and problem sovling, not just pattern recognition.

That said, consciousness is an incredibly heterogeneously defined term but can probably be pretty widely applied (many people confuse consciousness with sentience or sapience.)

That's a very good point - a key feature of the brain is that it's a highly dynamical system with a huge number of variables.

Neurons are all giving each-other feedback in real time, which leads to emergent behavior. To really reproduce this, you would probably need a very different hardware paradigm.

Quantum algorithms analogous to convolutional neural networks are O(log N) instead of O(N), so I'd wager we really do need something like that, or at least some other sort of analog hardware, given how complicated it is to do calculations on a QC.

Google has a few million hosts in it's datacenter. Each host has a few CPUs. Each CPU has a few billion transistors. That means there are on the order of 10s or 100s quadrillion transistors at Google. The human brain has around a quadrillion synapses. So, the rough numbers are there. Of course, that assumes the brain is like a computer, at all - my favorite hacker news comment someone made was the idea that maybe the computer analogy for the brain is incorrect and an analogy of an antenna is something that makes more sense. And a simulated antenna doesn't actually receive a signal ;)

A computer is a pretty bad analogy for a brain. But even so, 1 synapse is much much greater in complexity than 10 or 100 transistors. Each synapse is a complex system unto itself, which interacts with all kinds of cellular mechanisms which make up the highly dynamical system which dictates information processing within the neuron.

So if we wanted to make a simulation with reasonable fidelity to the mechanisms of cellular information processing we know about, I would wager to guess it's more on the order of 1GPU == 1Synapse at a minimum.

That's not to say that all of this complexity is required to reproduce consciousness in a meaningful way, but actual information processing in the brain is gob-smackingly complex.

We're the expression of consciousness; dipping itself into the world for a moment. The world we live in is made of the fabric of dreams; all our live is shrouded in eons of slumber.

Yeah, but a large portion of that is not involved in the higher level thought processes we care about when we talk about AI. AI doesn't have to micromanage a physical body.

Yeah, IMO if you want a rough idea about how much compute the human brain performs, and how efficiently you could do that in silicon, it's interesting to look at area V1 which is one of the lowest level areas involved in vision. This brain area comprises 140 million neurons per hemisphere (so roughly 280 million in total). So that's about 1/300 of the human brain.

The interesting thing here is, what V1 does is it computes directional receptive fields based on raw data from the retinas. We can implement that in a tiny silicon chip, maybe even just a DSP chip, not a full CPU. We know how to implement this kind of computation super efficiently, and without using neural networks.

We don't fully understand what goes on in many areas of the brain, but if we can find efficient ways to implement equivalent computations, we might technically already have sufficient manufacturing technology to implement equivalent functionality in a pretty compact and power-efficient form factor, or we might not be that far from there.

A classic quote from Dijkstra: "The question of whether machines can think is about as relevant as the question of whether submarines can swim."

I have my doubts about whether just scaling a model of V1 would be sufficient to reproduce consciousness. Structure and function are extremely closely linked in the nervous system, and neural circuitry is highly specialized in terms of structure to support a given function. E.g., motor cortex is going to be vastly different from auditory cortex and so on. Assuming that because we have 1st level feature detection in V1 fairly well understood and reproduced, we have solved neuroscience in general strikes me as a massive and unjustified leap of faith.

If there were a brain region that might be emulated to reproduce "general information processing", the best candidate would probably be neocortex. As far as I am aware, that does seem to be a very wide region of the brain which has fairly consistent structure, and largely serves to integrate all other systems.

No, scaling a model of V1 would not yield consciousness.

The point is that V1 is a large part of the visual cortex. It's something the brain dedicates a lot of resources to, and we could implement something similar in silicon almost trivially. Your smartphone GPU may have enough compute to simulate an equivalent computation at 200 frames per second.

It's possible that in the future, we'll be able to simulate consciousness very efficiently once we have a better handle on the kinds of computations that are involved, much more efficiently than what the brain does.

At what level of fidelity? 1 synapse == 1 16-bit floating point value? It may be enough (emphasis on may), for a specific brain region which is very directional in nature but I'm not sure if that calculation will generalize.

The brain is a pretty messy place. While our bodies try hard to preserve homeostasis, conditions will always vary slightly. Hormones affect our brains and can vary in concentration from moment to moment. Type 1 diabetics can easily have their blood sugar increase by a factor of 4, without losing consciousness. There are many other chemicals and conditions (temperature?) that similarly influence the condinions under which our neurons must operate.

But even in this noisy environment, our brains can usually operate just fine. Given this level of background noise, I would be very surprised if the average synapse would require more than a 16-bit floating point value for its outputs.

Counter argument:

Some synapses contain multiple receptor sub-types. Dopamine, for example, has some receptors which are inhibitory, and others which are excitatory. Each receptor is sensitive to different concentrations of dopamine. This gives the effect that when the dopamine concentrations in the synapse are low, this synapse acts to inhibit the activity of the efferent neuron, while when the concentration reaches a certain threshold, the output of the synapse flips to an excitatory signal.

How would you model that system with a single F16?

And that's just one example: for instance there are specific signals which trigger cascading effects in the neuron which up-regulate certain chemical pathways, causing an amplifying effect on other signals.

Or you have shunting effects, where an inhibitory synapse placed at a certain point in the dendritic arbor serves to selectively cancel signals from farther down that one specific branch of the dendrite.

That's not "noise" - the brain has incredibly refined and detailed mechanisms for information processing which go very far beyond `weight * sigmoid(activation) = output`

The body is just the brain's energy and survival system; biology is superficial

Why do you believe emulating a human brain is the only way to achieve consciousness?

Agree, are animals not sentient?

Animals are pretty clearly sentient, it is the sapience and consciousness that are harder to pin down.