AI as a Tool for Energy Abundance and Safety
How today's AI is improving the energy sector
Image source: Wired Magazine.
The United States needs more energy. With our energy needs growing rapidly, especially in the computing sector, we need more efficient methods of producing, storing, and transmitting electricity. Artificial intelligence (AI) is already playing a pivotal role in making energy systems safer, more efficient, and more innovative.
Here are three ways AI is on the job, tackling the energy challenge.
First: AI Can Help Make Nuclear Power Safer and Cheaper
Consider Northern Virginia: its 442+ data centers, the largest concentration in the world, handles an astounding 70% of global internet traffic. These data centers consume vast amounts of power, and as more are built, energy demands will soar. Some estimates predict that energy demand driven by northern Virginia’s data centers could increase by 85% in 15 years. Utility companies are feeling the heat, and along with data center developers, are calling for more carbon-free energy. Increasingly, these developers see nuclear power as a crucial part of the solution.
To many, the word “nuclear” triggers concern and even fear. Nuclear plants are highly complex systems of interlocking sensors and machinery and often rely on imperfect human judgment for crucial operating decisions. It’s widely understood that the more complex a system is, the greater the risk of system failure. Human error is cause for concern.
Thankfully, AI can help. Two years ago, the International Atomic Energy Agency (IAEA) outlined “Seven Ways AI Will Change Nuclear Science and Technology.”1 Specifically, the IAEA highlights that AI stands to strengthen and promote peaceful applications of nuclear technology by using its power to monitor plant performance and detect anomalies before they cause trouble.
A particularly promising use of AI in nuclear power is predictive analytics. By creating “digital twins” (virtual models) of nuclear plant systems, AI can simultaneously monitor and analyze scores of data streams produced by generators, detecting anomalies, predicting maintenance needs, and preemptively addressing safety issues. Safeguards like these could make nuclear energy a more stable, cost-effective option. Argonne National Laboratory reports that nuclear operation and maintenance account for 70% of operating expenses for the nuclear industry. Intelligent predictive maintenance could minimize those budget burdens. The potential result: increased safety paired with potentially significant cost reductions needed to put the industry on stable footing.
Second: AI can Improve Energy Industry Efficiency
The energy sector’s biggest challenge is maintaining “just in time” alignment between energy supply and power demand. A mismatch between the two results in wasted carbon and higher costs. AI-enabled energy distribution systems can improve efficiency by synthesizing data from generators, weather prediction systems, smart meters, and other sources to more precisely balance supply with demand. These systems already manage around 4% to 8% of U.S. energy capacity, and their use is growing. Studies show that such advanced AI technology could yield significant cost savings in the energy industry – potentially hundreds or thousands of dollars per consumer – while reducing carbon emissions and increasing overall stability of the grid.
Third: AI Can Be Used to Discover Critical Energy Materials
AI is helping scientists locate and utilize minerals essential to energy technology. Researchers recently used AI to reveal substantial lithium deposits within the Smackover Formation in southern Arkansas. This geological remnant of an ancient sea contains brine—salty water—rich in lithium, a key component of batteries used to power a wide range of devices, including electric vehicles. Combining AI and machine learning techniques with water testing, researchers from the U.S. Geological Survey (USGS) estimated that the brine in this formation holds between 5 and 19 million tons of lithium, potentially sufficient to meet global demand for electric vehicle batteries many times over, accelerating industrial scale energy storage within years instead of decades.
In addition to finding new sources of minerals, AI systems are helping scientists discover new and materials for energy production and storage. Studies show that AI-assisted materials research has already proven itself, resulting in new and better materials for solar energy technology and greater downstream product innovation.
By making nuclear power safer, improving the efficiency of energy distribution, and accelerating the discovery of new materials, AI is helping to create a more abundant and sustainable energy future. Let’s keep the momentum going!
Written In collaboration with Dr. Christine McDaniel. Special thanks to Mark Ingabretsen for helping with editing.
(These are discussed in more detail in the IAEA publication Artificial Intelligence for Accelerating Nuclear Applications, Science and Technology
Letting “divorce-your-wife” Bing, “humans please just die” Gemini & co run nuclear power plants… hmm, what could possibly go wrong? And I thought our worst case scenario was that one of these AI systems could hack into a nuclear power plant. Def better to give them full access right away!