BlocksInform
Sam Altman, the CEO of ChatGPT creator OpenAI, has recently started a new conversation regarding nuclear fusion by stating that it could be the solution to artificial intelligence’s thirst for electricity.
As per Atlman, nuclear fusion will allow the generation of vast amounts of power without the worry of carbon footprint – which will be the biggest boost for AI.
The OpenAI CEO made these comments while speaking to Lex Fridman over a podcast interview.
However, it is also important to mention Altman himself has invested millions of dollars in nuclear energy.
What did Altman say?
Altman said that the next generation of AI will reach a level that surpasses human intelligence. According to him, AI will require vast amounts of energy to operate, and nuclear fusion would be the best bet for its source.
In an interview in January, Altman had stated, “There’s no way to get there without a breakthrough, we need fusion.” He reiterated the same in the March interview given to Fridman.
Scientists have repeatedly pointed out that despite breakthroughs, the goal of generating power through fusion on a wide scale is far away.
During the interview, Altman stated that sourcing energy for the growing AI needs is the most challenging part.
The Open AI CEO has invested $375 million in the US-based nuclear fusion firm Helion Energy. The company has set a target of producing energy at a commercial scale by 2028.
“I think Helion’s doing the best work,” Altman said during the interview. “But I’m happy there’s a race for fusion right now.”
During the interview he had also pointed out that artificial general intelligence (AGI) might be a reality in this decade. He had added that building AGI would be a new beginning.
Problems due to AI
While AI has been gaining popularity, with new versions being released every month by various companies, there is also a growing concern about its disproportionate usage of natural resources and energy.
AI use is being linked with carbon emissions from non-renewable sources of electricity and for its huge consumption of fresh water.
Large AI models are often trained and deployed on large clusters of servers with multiple graphic processing units (GPUs). These servers can consume a few kilowatts each, the equivalent of the average power consumption of an entire house.
AI models also use fresh water in two ways: onsite server cooling and offsite electricity generation. The mega energy use by AI servers generates a lot of heat, and so they need water-aided cooling towers and/or cool air to avoid server overheating, which makes it a case of onsite water consumption.
Moreover, electricity generation also uses up a lot of water. AI, therefore, is also responsible for offsite water consumption.
Nuclear fusion, once achieved at scale, could provide a cleaner and more feasible solution to the energy problem for AI models.
What is nuclear fusion?
Inside a fusion reactor, isotopes of hydrogen are heated to extremely high temperatures and pressures to form superheated plasma. The intense heat and pressure causes atomic nuclei to collide and fuse, releasing vast amounts of energy.
This process is how the Sun and the stars shine; however, generating controlled fusion energy on Earth has been a persistent challenge.
One of the persistent challenges is the inherent stability of the superheated plasma. Moreover, there is also a great deal of apprehension regarding fusion energy, partly due to misconceptions about its safety.
On the plus side, fusion reactors produce minimal long-lived radioactive waste compared to traditional fission reactors. This reduces the challenges associated with nuclear waste disposal.
