InnovationResearchScience

Google Machine Learning Tech Aids Nuclear Fusion Research

As News Editor of Silicon UK, Roland keeps a keen eye on the daily tech news coverage for the site, while also focusing on stories around cyber security, public sector IT, innovation, AI, and gadgets.

Follow on:

Working with Tri Alpha Energy, Google has developed the research accelerating Optometrist algorithm

Google and nuclear fusion company Tri Alpha Energy have developed a machine learning algorithm to accelerate experiments on turning plasma into energy. 

For the past three years the two companies have been working together to advance research into plasma, which could lead to the development of nuclear fusion power, a cleaner and safer alternative to nuclear fission, that has the potential to yield more power.

But is extremely complicated to achieve, with fusion experiments consuming more energy than what could potentially be yielded, and the ability to harness energy from fusion is a controlled and impactfull matter has yet to be achieved. 

Even if that were not the case,  fusion has yet be done at scale in the same fashion as the fission found in nuclear power plants. 

But Google and Tri Alpha Energy have come up with the Optometrist Algorithm to enable a combination of powerful computation augmented by human knowledge and judgement to speed up fusion-centric plasma experiments. 

Fusing machine learning with human smarts 

Fusion machineIn basic terms, the algorithms works by presenting human experts with successive pairs of possible outcomes from an an experiment, enabling the expert to make a judgement on the data served up and steer the direction of following experiments. 

While Google has done lot of work with machine learning that can tap into the compute power of its data centres and Google Cloud Platform, Ted Baltz, senior staff software engineer at the Google Accelerated Science team, noted that sifting through all the data and carrying out analysis on plasma research is beyond the capabilities of even Google’s massive compute resources. 

“The whole thing is beyond what we know how to do even with Google-scale computer resources,” he said. 

“So we boil the problem down to let’s find plasma behaviors that an expert human plasma physicist thinks are interesting, and let’s not break the machine when we’re doing it.” 

As a result of this approach, Tri Alpha Energy has seen its experiments speed up dramatically, reducing operations that previously took months into a few hours. And the Optometrist algorithm also helped the research team achieve a 50 percent reduction in the energy losses from its C2-U giant ionised plasma machine, which lead to an increase in total plasma energy, a core part in reaching the threshold for fusion happen. 

The next steps in the Google and Tri Alpha Energy will involve applying the  Optometrist algorithm to a new plasma machine called Norman, with the aim of further making breakthroughs in fusion research. 

While there are many who fear that machine learning and the development of artificial intelligence will put jobs at risk and even threaten the existence of humanity, in this case we see a smart technology being put to use in a way that benefits people rather than replace them. 

Put your knowledge of artificial intelligence to the test. Try our quiz!