What if we could move computer tasks around the world to make better use of renewable energy? I talked to Elena Vindrola about how AI could make this possible. We also discussed the environmental impact of AI, how we can measure its net gains, future technology, and how youth today can make an impact. Continue reading to learn more about Elena :)

Elena Vindrola is a PhD student in computer science and engineering at UC Santa Cruz, where she researches sustainability for data centers at the Sustainable Systems Lab. The lab studies how to schedule computer tasks to reduce carbon emissions using power grid data. Specifically, Elena uses machine learning to understand how jobs consume energy and how they can be scheduled to reduce spikes in electricity demand, which can strain the grid.

One particularly interesting idea we discussed was moving workloads between data centers to take advantage of renewable energy. Renewable energy isn’t always reliable because its availability varies throughout the day. For example, the sun isn’t always shining, so solar energy cannot always be generated. Elena explained that she wants to explore shifting workloads between data centers to optimize renewable energy use. If a workload comes in at a California data center during the night, for example, it could potentially be shifted to an Australian data center where it is daytime, and vice versa.

One way to think about moving workloads is that it isn’t like a washing machine in your house, where you can only wash your clothes in one place. Instead, you can move the workload where it’s more convenient. By moving computer tasks to locations where renewable energy is more available, data centers could potentially reduce their reliance on fossil fuels.

I think this is a very cool idea because it shows how AI and computer science can help solve environmental challenges in ways that aren’t immediately obvious. I’m excited to see how this idea could be implemented in the future!

Another question I asked Elena was how we can measure the benefits of using AI for sustainability when AI itself can harm the environment. She explained that we need to estimate the long term cost benefit over 25-50 years. For example, we could compare how much energy a grid optimization model saves each year to how much energy it consumes. Like installing solar panels, the benefits may take time to outweigh the initial costs. Elena emphasized the importance of working toward a positive net gain by having power system and AI experts collaborate to find the right solutions.

We also discussed another challenge for the power grid: connecting millions of EVs (electric vehicles) to the current power distribution system, which can create spikes in electricity demand. Elena explained that researchers are investigating how EV batteries could make the grid more resilient by supplying energy back to the grid when needed. For example, people could get paid for sharing their car’s battery power when they aren’t using their vehicles. However, one concern is that frequently charging and discharging batteries could shorten their lifespan.

Lastly, we discussed how students interested in AI and technology can make an impact. Elena encouraged students to identify problems in their communities and think creatively about how they can solve them.

Thank you so much, Elena, for having this discussion with me! I really enjoyed learning about how AI could make our energy systems more sustainable, especially through the idea of moving workloads between data centers. I hope readers enjoyed it as much as I did!!