A few weeks ago, my youngest daughter was sick with the flu — she had a fever for a full week, which was concerning but not panic-inducing. But my two older daughters, (aka, moms two and three to their little sister), turned to Google to learn all about fevers. They came running to me with the knowledge that Google’s AI summary said a 104° F fever is cause to rush to the hospital. But it wasn’t my first rodeo, so I showed them dozens of legitimate hospital and medical expert websites, all stating the exact opposite.
So where did this AI summary come from? Literally every resource right below it said 104° F fevers are not a concern, but it still summarized that incorrectly. This got me thinking — just how much should we be using AI in our jobs? Not only is it inaccurate at times, it consumes more energy than most users realize. According to the World Economic Forum, Microsoft recently announced its CO2 emissions had risen nearly 30% since 2020 due to data center expansion. Similarly, Google’s GHG emissions in 2023 were almost 50% higher than in 2019, largely due to the energy demand tied to data centers.1
To understand the value of AI, I dug deeper to learn how manufacturers are using it. One friend who owns a local manufacturing facility uses AI to do research for his current customer base. His team found that AI often has small but important technical and mathematical errors, so it’s critical to go back and check its work. He added that sometimes, a number might be off by a decimal point, or its source information may be inaccurate. He compared it thus, “Some days you can have graduate assistant level work, and then other days it’s more in line with a summer intern with good grades and a resume who landed their first real summer job.” For him, it has value, but it definitely requires a good review before you say the job is done.
I also learned from a colleague how the company Monumo is redesigning electric motors with deep tech engineering. Its machine learning-friendly Anser Engine can now run 10 million simulations in 24 hours and identify the optimal parameters for specific motor use-cases to reduce cost, increase efficiency or improve sustainability.
These are both great examples of how we can use this still very new and constantly evolving technology to better our systems, machines and components. Here, AI is assisting engineers and technical personnel to design the machines of the future and benefit manufacturing.
Considering energy use (and accompanying emissions), it’s wise for manufacturers who want to use AI to do extensive research and find applications that benefit from it best. Like my personal experience, Google search was already doing a better job of answering questions with accurate resources. Perhaps it’s time to put that type of AI behind us and focus our energy (literally and figuratively) on applications like these here to truly better engineering.
1 https://www.weforum.org/stories/2024/07/generative-ai-energy-emissions/
Filed Under: Trending
