‘Robots will take our jobs’, so go the headlines as Artificial Intelligence flexes its muscles ever more impressively and yet again we debate the coming or not of machine consciousness.
In the 1980’s Roger Penrose1 first concluded that ‘mind’ or intelligence was not a result of processes analogous to digital computing. Forty years on, despite the astonishing achievements of computing he remains of that opinion. Somewhere, he and others believe, the answer to mind lays in the quantum world replete with its mysterious world of probability, superposition and entaglement.
Today2, his search is for biological mechanisms and structures that could support such quantum states and put clear blue conceptual water between man and machine . Ironically at the same time, in the hitherto deterministic digital world of computing, machines are being constructed that use the same quantum principles to create probabilistic computational devices based on ‘qubits’.
For a biologist these directions of travel look like another example of convergent evolution, viz the similarities between sharks and dolphins or birds and bats*. Convergence may be the key to understanding the gulf between AI /robotics and the mind of living creatures. I think we should see it this way:
The biological mind allowed creatures to behave like robots.
Robots increasingly are able to replicate our ‘robotic’ behaviours ...
To be more specific, a creature behaves ‘robotically’ when it carries out simple or complex algorithms in response to its environment. For example, the move towards food or light by a simple organism is robotic and readily replicated by a robot equipped with appropriate sensors and algorithms.But ditto more complex behaviours such as car assembly, playing poker, Go!, chess, playing the stock market or diagnosing illness by humans can be replicated or even bettered by robots so long as the behaviour can be represented algorithmically. We should not then be surprised when robots out-robot us! But how can we behave like robots?
The ability to do this comes with the appearance of the eukaryotes. Environmental responses by simpler organisms are known ( taxis along chemical gradients) but rare.
Eukaryotes are by contrast very complex they have a nucleus, mitochondria, and cytoskeletons. The DNA of the nucleus with its store of genes amounting to a vast permanent library of information, tempts us strongly to think in terms of familiar digital storage metaphors such as ‘hard-drives’ and ‘files’. Seductively the nervous system, especially the brain, is a shoe-in for the modern computer’s CPU as first envisaged by Von Neumann in the 1940s. Together they create the computer analogy rejected by Penrose.
But single celled creatures without a nervous system have minds too! The advanced behaviours of single-celled Dictyostelium discoideum or ‘slime mold’ is a good starting point. It can transit between motile hunting amoebae to aggregated, mold-like multicellular creatures. No nervous system is available to act as a classical computational device, a CPU, so what is?
There are two candidates in my view. Penrose favours the cytoskeleton and asserts that this structure is able to maintain coherent quantum states. My mitochondrial chauvinism disposes me to favour the mitochondrion for the following reasons:
- Mitochondria separate electrons from protons and are able to move electrons through a redox chain of proteins using quantum tunneling in the process known as oxidative phosphorylation. Ultimately hydrogen and electrons are united with atomic oxygen radicals to form water.
- The mitochondrial membrane is rich in cholesterol and hydrophobic lipids and so has a high dielectric which is uses to maintain a constant charge separation known as the membrane potential.
- Each mitochondrion has thousands of these supercomplexes of respiratory ‘chains’
- Mitochondria can hold and trap ions within their multi-protein super-complexes
In mitochondria we have permanently charged devices capable of thousands of quantum separations each with the possibilities known as entanglement and superposition.
Meanwhile (to fuel my fantasy further), in the digital world promising qubit implementations have been made using the trapped ion quantum3 states.
How would a mitochondrial qubit machine work? Well how does a digital qubit machine work?
Basically they would work in the same way. That is qubits would be in a superpositional state until an external interaction causes this to collapse into a regular state. So for example, in Computing, a ‘bit’ can exist in two states 0 or 1 (or any binary state that is equivalent). A qubit though can be 0 or 1, or both at the same time until an interaction causes it to ‘collapse’ into either 0 or 1. You will have to trust me that these computers do work though the output is not determined absolutely as in the non-quantum world, it is expressed as the most probable outcome.
Enough of quantum computers, a little knowledge is enough to make it conceivable that a similar indeterminate superpositional state of affairs held by mitochondria could collapse into alternative fixed states as a result of environmental stimuli. A nervous system, a brain indeed, is simply an elaboration of the basic setup.Here we have cells that specialise in quantum computing, rich in mitochondria and responding to simple electrical stimuli rather than the full spectrum of the outside world.
In summary we should not be surprised that the external lives or behaviours of computers and creatures are on a converging path. It may be that their underlying technologies are also converging, but of their inner lives there is not much we can say other than they have them; like birds and bats they will be similar but utterly unalike.
*Dolphins and bats are mammals which appeared about 50 million years ago whereas sharks and birds appear over 400 millions years ago.
- The Emperor’s New Mind by Roger Penrose 1989 OUP ISBN 0-19-851973
- Shadows of the Mind by Roger Penrose 1994 OUP ISBN 978-0679454434
- https://en.wikipedia.org/wiki/Trapped_ion_quantum_computer
No comments:
Post a Comment