I am somewhat dubious of the claims made in the article. The video is very good though, although at the very start he repeats a few common tropes about seawater being the fuel for fusion and that fusion is "clean". I work at ITER, so have a vague idea of the state of things...
Fusion is not exactly fuelled by sea-water. He is referring to Deuterium-Tritium fusion reactions. It is true that deuterium can be extracted from seawater, but the other isotope of hydrogen required, tritium, is only available in trace amounts in the environment and is a fast decaying radio-isotope (half-life 12.3yrs). There is only ~3.5kg of tritium naturally available at any one time on the planet. Instead, the required tritium must be bred inside fission or fusion reactors by neutron activation of lithium.
Fusion is also not clean. The neutrons produced in a fusion reaction activate elements in the materials of the machine to create radioactive isotopes. E.g. Cobalt-59 in stainless steel is activated to Cobalt-60 which is a gamma emitter with a half-life of around 6 years. Although we won't create High-Level Radioactive waste, we will produce vast quantities of Medium and Low-Level Waste which must be handled as radwaste and which still needs to be packaged and stored for several years.
One problem with articles such as these is that when scientists say XXX will be possible in YYY years time, the scientists often have not given much consideration to the future engineering challenges to be faced in actually realising the theory. A big problem with fusion is the time frames people have put on what is an incredibly complex challenge. It has always raised unrealistic expectations.
It is certainly interesting to see modern computing approaches helping, but in reality, if AI could solve it all then we have just wiped out millions of jobs across the world.... AI cannot solve all engineering design challenges. Engineering is a combination of creativity and logic. AI, I suppose, is cold, hard logic. There are thousands of engineers at ITER dedicated to design and analysis to support the science with perhaps tens of thousands of highly-specialist technicians and operators in the supply chain. AI would also not solve supply chain issues.
The major issues related to realising fusion, at least in Tokamaks are:
- maintaining a stable plasma - physics
- solving novel material science and engineering challenges to ensure that the machine doesn't just disintegrate and clog itself up with dust created from particles stripped-away from plasma-facing components by the neutron flux
- solving the very complex challenges of remotely-operated handling and maintenance operations inside the vacuum vessel, which are necessary just to keep such a machine running
- the huge material and engineering costs
- breeding sufficient amounts of the required tritium to feed the machine
- treating the radwaste
All of this is on top of the fact that a fusion reactor is a nuclear installation and must therefore be licensed by the state agency responsible for nuclear safety. This itself is a huge challenge, particularly for a novel plant like a fusion reactor. "Beginning commercialisation" in 2023 I guess means only just starting the design approval for the reactor, which will take years.
Very few of the above challenges could be solved by AI.
I am confident that we will see a commercial fusion reactor connected to the grid in our lifetime, but it won't be ITER and it won't even be the follow-up, DEMO.
ITER itself will not generate any electricity, it won't even have a turbine island, it is an experimental reactor. Instead, the purpose is to generate a plasma with fusion power ten times the external heating power applied to the plasma. ITER is essentially a proving ground for the plasma physics, materials science, and engineering solutions applied to the magnetic confinement approach to fusion in a Tokamak machine. At sometime during the life of ITER the member states will begin to construct their own demonstration (DEMO) reactor designs, which will be followed up by commercial reactors.
In any case, nobody I know at ITER would be upset if another organisation suddenly cracked fusion and got it to work. Everybody is working toward the same goal and solving the challenge can only be a good thing for humanity. Although personally, I believe it's probably better for everybody if it was cracked by an international collaboration than an investor-funded private entity...
Elusive, not illusive.
