Is AI a climate change carbon bomb, or the path to sustainability?

New artificial intelligence algorithms could accelerate the energy consumption crisis, warns VU Amsterdam climate researcher Alex de Vries.

Data scientist and climate researcher at VU Amsterdam, Alex de Vries, warns of a potential energy crisis driven by the enormous energy demands of high-performance AI models. The scale of energy usage could surpass high-energy industries like cryptocurrency mining if the tech industry quickly adopts AI.

“The impact of AI on our global energy consumption depends largely on the pace of AI adoption. If [the tech industry] wants to push AI [harder] in the next couple of months, we will have a big problem," said de Vries in a recent interview.

This energy consumption primarily arises during the continuous deployment of AI models, which is more energy-intensive than the one-time training process. Tech giants such as Google, OpenAI, Amazon, Apple, and Meta are major players in the AI revolution.

Even a small fraction of AI investment by these companies could lead to energy consumption exceeding that of Bitcoin miners. The situation is complicated further by the unpredictable AI adoption curve; an overnight demand surge might outpace the supporting technology and machinery, causing energy bottlenecks. Global energy consumption could spiral out of control if significant resources are spent on resolving these bottlenecks.

Importantly, de Vries notes that AI isn't just a potential problem; it's also part of the solution to climate change. AI can optimize energy usage, develop more efficient renewable energy sources, and improve climate prediction models. By identifying patterns and predicting outcomes, AI can streamline energy use, foster the creation of smarter and more efficient renewable energy technologies, and refine climate models, contributing significantly to the fight against climate change.