In combination, the individual paradigms of AI, fusion and quantum could change the trajectory of history. The timescales are now more certain, as milestones are being met. This has released more capital for research, investment and actual builds. It now seems possible in the 2030-40s that this holy trinity could save us, as well as give unimaginable leaps in productivity.
Multipliers
There are several multipliers with technology that are often missing when we see technology as linear progress of a singular technologies; writing, printing, steam, oil, electricity, radio, TV, personal computer internet, smartphone and now AI. They are often presented as a series on a graph. This linear thinking misses combinatorial technology.
Technology begets technology
The iPhone was truly combinatorial as was the printing press. Gutenberg did not ‘invent’ printing. He brought together several different technologies that had crossed usable thresholds. Movable Type, which came from China and Korea, relied on metal alloy development using lead, tin and antimony, along with new casting techniques and precision moulds. The press was adapted from wine/olive presses with a screw mechanism for applying even pressure. Then there was paper technology, brought from China, using European paper mills that has designed manufacturing processes based on linen-based paper production, far more economical than parchment. And don’t forget the new oil-based ink, far more durable than water-based inks, with better adhesion to metal type, that lasted longer on paper. Finally, the need for socially agreed standardisation on type; letters, punctuation, capitalisation, agreed standardisation of type sizes and consistent letter heights. It was the combination of all of these elements, some taking centuries to develop, others quick spurts, that came together in a systematic way to allow the mass production with efficient workflows that created printing in the 15th and 16th centuries. This combinatory technology had a profound effect on our culture, politics, the trajectory of history, as well as unleashing our ability to create new technologies.
Will AI hit walls of energy & compute?
Other technologies can combine with other gargantuan leaps to make exponential progress – with much bigger and faster leaps. Some technologies, like micro-processors, have intrinsic exponential progress (Moore’s Law), as they got better and faster. There are all sorts of laws that define exponential growth across networks. But one thing that halts growth is the sudden collapse of a technology as it hits a major constraint.
With AI continuing its exponential ascent, many see it hitting the brick walls of energy and/or compute. Yet, a push on one technology often results in the pull of another. We see Google, Amazon and others invest in Fusion. Google and others have also strongly invested in Quantum computing. This can be seen as the Holy Grail of contemporary tech – AI, Fusion and Quantum.
Fusion
Emission free energy is a constraint on AI. A stepping stone to Fusion, practically, not technologically, are SMR nuclear reactors. That’s solves the emissions problem. It does not necessarily solve the energy demand problem.
Fusion comes, not from splitting the atom but fusing atoms – fusion. Push Hydrogen atoms together to form helium and huge amounts of energy are released. Target gain was achieved recently (more energy produced than the amount required to perform the experiment) and they now have the brains and money to make fusion happen.
Major tech companies have invested in Fusion, which promises unlimited energy at lower cost. Microsoft have a major investment in Commonwealth Fusion Systems and have partnered with ITER for computing support. They are keen on providing computer support for start-ups and have several hundred million invested. Google invested in TAE Technologies, who use AI for plasma control research and computational support for fusion modelling. They also have a partnership with Princeton Plasma Physics Lab. Bezos Expeditions invested in General Fusion and AWS is also providing computing resources to fusion projects. Amazon itself has made multiple rounds of funding into Commonwealth Fusion. Eric Lander led the Human Genome Project has a MacArthur ‘genius’ grant is also head of the head of the US Office of Science and Technology Policy for Biden. Recent breakthroughs have led him to become CEO of Pacific Fusion.
Quantum leap
Conventional computing is a constraint on AI. Quantum computing promises to do in minutes what took thousands of years to solve problems. It changes the whole idea of what is computable. Once again, like Deepmind’s Alphafold saving centuries of research with one piece of AI software, we have a technology (quantum computing) that could accelerate another technology’s (AI) capabilities. This promises enormous productivity gains in research and problem solving.
AI could be accelerated with quantum neural networks, better pattern recognition and much faster solving of optimisation problems. Climate change could be modelled with weather prediction, climate change simulations, as well as atmospheric modelling and ocean current analysis. Our energy production and grid management could also be far better managed with power flow optimisation, load balancing, grid stability analysis and better renewable energy integration. This reduces the load needed for global use of AI.
Leaps in productivity include cryptography and security, to keep personal and other data safe through quantum key distribution. Drug discovery and materials science through simulating molecular interactions and designing new materials with specific properties, for example, in battery development. Supply chain and logistics could use route optimisation across delivery networks, warehouse optimisation and fleet management.
It can also help solve several critical challenges in both fusion and nuclear reactor design. In fusion on plasma physics simulation and materials science, also modelling in fusion. Small Nuclear Reactors (SMRs) may also benefit from modelling and materials science. AI solves problems, Fusion fuels AI. It is a virtuous circle.
Quantum computers still have limited qubits and they still have high error rates that require error correction but they exist and AI companies like Google have significant investments in the field. Incredibly difficult to build, they require super-cooling to temperatures unknown to us in the past. They are also sensitive to environmental interference. This raises the problem of cost and access to quantum hardware. But progress has been made.
Quantum computing could also potentially help solve several critical challenges in both fusion and nuclear reactor design.
Conclusion
In combination, the individual paradigms of AI, fusion and quantum will change the trajectory of history. The timescales are now becoming more certain, as milestones are being met. This releases more capital for research and investment. It now seems possible in the 2030s and 2040s, this holy trinity is likely to save us, as well as give unimaginable leaps in productivity.
Note that AI is here. That's a given. Fusion is a known process (Sun is one giant Fusion reactor burnitruning 4.5 million tons of mass to energy a second) and Quantum computing (known in physics) has been built. This is not blue-sky research on theory but design and engineering problems. Those can be solved. Time will tell.
No comments:
Post a Comment