Transhumanism Part 2.5. Quantum Computing & the Brain

In part 1, we briefly exposed some of the dangers related to AI. Cohabitation is a possibility but seems somewhat improbable unless humanity finds a new function to fill in the new "natural" order. In the next 15-25 years, we are going to see the creation of Quantum supercomputer. The basic Quantum Mechanics concepts have been exposed in Part 2 of this series. Those computers are able to calculate in term of fuzzy logic. Meaning that they no longer do the 1 and 0 of a classical Turing-style computer (Boolean logic), but rather work in odds. Odds of being 1 and odds of being zero. Not only that but they are at any time every single possible answer [see the famous Schrodinger Cat]. Difficult to grasp. Don't worry, Albert Einstein himself was creeped out by some aspects of Quantum Physics. And we know much more about it nowadays than during its time. I will probably write something about quantum computing in the future. Long story short, it is very similar in its inner logic to the human brain which is also a fuzzy logic biological machine. We can safely imagine that one of the outcomes of having such an incredible machine is that we are going to leap forward in term of Artificial Intelligence development. 

The human brain is currently the wiring of more or less 86 billion neurones [The Human Brain in Numbers: A Linearly Scaled-up Primate Brain, Suzana Herculano-Houzel, 2009 Nov 9]. Each neurone can be connected to 100 000 other cells trough the dendrites, hence our brain is an incredibly complex machine of up to 8.6 thousand trillions of connections. To put this in perspective with the current state of the technology, an 18-cores Xeon Haswell-E5 Intel CPU chip got 5.5 billion transistors. In addition, those transistors are far from being well connected to one another. The connectivity between them is much less efficient than any biological brain. To give an idea of the scale, if we wanted to create a super CPU with as many transistors as there are neurons in the human brain we would have to connect 15 of such CPUs, and even then this would not solve the interconnectivity issue as we would need to connect each transistor to 100 000 others (As of now, the fan-out number is more about 4-6). We can see that we are very far from the multiplicative factor. Now this is in term of raw numbers, but what about the speed? Well, an action potential running trough the axiom of a neurone is going roughly at an average of 25m/s (but can go from 1m/s to 100m/s in hyper-myelination cases), so the propagation of signal speed is almost irrelevant. The neurones can shoot an impulse (action potential) every 5ms when well rested. This means, making the false assumption that the neurones can maintain such pace, that it is working at a frequency of 200 impulses per second or 200 Hz.

We can compare this to the silicon CPU which it is working on a clock of 2.3 GHz. This is more than 11 million times faster. We can hence make an interesting postulate. What makes the power of our brain is not really its speed. It is indeed relatively slow. It is its interconnection, aka its complexity as we will show later. As you can see, it would be almost impossible to bring such an incredible amount of connections using the actual silicon manufacturing paradigm (We are already at a scale of 22 nm, and a single atom goes from 0.1 to 0.5 nm). 

So what is the link with quantum computing? Quantum computing can solve this because each qubit is linked to one another by Quantum Entanglement and can exchange information instantly. [Harnessing high-dimensional hyperentanglement through a biphoton frequency comb." Nature Photonics (2015) ]. Hence the number of connection is super-exponential, making the number of connection of the brain reachable, potentially artificially replicating the complexity of our brain and even beyond as we will see in a further section. For the computer literate, it can solve non-polynomial problem in polynomial time. What is the feasibility of such idea? Well, one of the most advanced laboratory in the world, the QuAIL (Quantum Artificial Intelligence Laboratory) from NASA is currently conducting research in this direction in partnership with Google/Alphabet and the Universities Space Research Association, and the Centre for Quantum Computation and Intelligent Systems (QCIS) from UTS is actually developing the software language which might be needed for such enterprise, but the race is also on with IBM, Microsoft, and DARPA. 

In the next part, we will go deeper into the quantum part and its link to biology.