Dr. Ulrich Spiedel from the University of Auckland has proposed a hypothesis that artificial intelligence could be the key to overcoming the environmental challenges it helps provoke, including its colossal energy consumption.
What Happened
The scientist proposed a concept of a "closed-loop self-optimization cycle," in which AI algorithms are used to accelerate research in materials science and energy. Specifically, AI can optimize the performance of solar panels, accelerate the search for new chemical compositions for batteries through simulations, and design next-generation energy-efficient semiconductors.
Context
According to a UN report, by 2030, the energy consumption of AI systems could reach 3% of global electricity levels. This creates an urgent need to find ways to reduce the load on data center infrastructure.
Why It Matters for the Industry
The industry is transitioning from a stage of pure resource consumption to a stage of actively seeking ways to optimize infrastructure. Using AI to accelerate R&D in materials science and chip design could radically reduce the cost and energy requirements of scalable computing.
Why It Matters for Users
Technological progress could create a cycle where algorithms help build more efficient hardware and energy sources for themselves, which in the long term could slow the growth rate of energy consumption and potentially lower the cost of computing power.
What Is Not Yet Known / Limitations
At the current stage, the concept of a self-optimizing cycle remains a theoretical hypothesis, and perceptions of its feasibility vary depending on professional roles: ranging from researcher skepticism to product developer optimism.
Sources
Author
Look at AI, Editorial