Many of us at Napa Vision 2050, and we are sure many of our Napa neighbors, are concerned about AI (artificial intelligence). It’s what everyone is talking about — struggling to understand it and learning about its power — yet we know it poses significant environmental impacts. It comes to us under many names: Claude, ChatGPT, Gemini and others.

 

Why do I care about this issue? I have two UC-educated sons, engineers working in technology, and grandchildren, whose future is at risk from climate change. And like all of us in this beautiful valley, I care deeply about its viable future.

 

The infrastructure of AI requires massive data centers on large properties that could be sensitive wildlife habitats, uses large amounts of water to cool its machinery and draws vast amounts of electricity. All of this concerns us. No one wants it here.

 

AI is utterly seductive. Using AI can feel like magic, even like a new best friend. We can summon it with a click on our laptops or a voice command into our phones. It can solve the simplest problems, like what can I cook with rhubarb? Or where should we go with three days in Monterey? And why does my shoulder hurt so much? But that is just the tip of the iceberg.

 

AI performs very complex math and does it fast. It aggregates data and supplies information that already exists in many disparate places. Even a very smart human would take many months to pull together such information. The truth is that we don’t have time to muddle. With the accelerating pace of climate change, we need answers and solutions now.

 

To successfully navigate our ecological future and protect our measurable resources, it is imperative that we look beyond this dichotomy and understand how this technology’s heavy resource consumption intersects with its ability to solve the very crises it exacerbates.

 

Yes, it feels jarring and ironic to look to a water-thirsty technology to study or solve a water crisis. Yet it is increasingly evident that AI is both a massive ecological strain and one of our most potent tools for environmental stewardship.

 

Let’s look at some good results. We are witnessing this work in platforms like Water Audit California’s new data portal. I talked with Bill McKinnon, the lead attorney for Water Audit California, about this. He explained that “streams are public trust resources, and no one can protect what isn’t measured. Every new well, diversion and reservoir release changes how much water is in a watercourse, yet those changes usually go unrecorded or reconciled with environmental needs. Water Audit’s dashboard reports conditions so everyone can see when a stream is under stress.”

 

Research is backing up the need for this data. McKinnon adds: “UC Berkeley researchers found that only about 8% of California’s rivers and streams have stream gauges. As part of our mission Water Audit California is filling that gap.”

See for yourself at data.waterauditca.org, now showing real-time readings from public and private gauges in the Napa River watershed, Putah Creek and Sonoma County’s Dry Creek, together with groundwater extraction and well-level data.

 

Managing finite water supplies across complex watersheds, agricultural sectors, rivers, streams and growing municipalities requires processing mountains of fragmented measurements. AI can analyze everything from stream gauges, flow rates, depth and snowpack measurements to satellite-based land tracking. This is where machine learning leaves humans in the dust. We need speed and accuracy more than ever.

 

Think of the forecasting that is possible. AI can analyze complex water data and use its algorithms to model groundwater depletion rates, predict drought severity and forecast streamflow fluctuations, dry conditions and even potential flooding, with unprecedented speed.

 

Platforms that aggregate this data help California counties fulfill their groundwater sustainability plans. Local residents and nonprofits seeking to protect our water resources can monitor trends. We can track, measure and manage scarcity before it cascades into an irreversible crisis. In this context, data systems are tangible tools for civic and agricultural survival, using computing power to protect our ecosystems.

 

Ultimately, the future of AI will not be decided by pretending its environmental footprint does not exist, nor by panicking and rejecting its potential. The path forward demands strict accountability, carefully chosen locations for data centers, demanding renewable energy transitions for server farms and, very importantly, establishing water-sparing, closed-loop cooling systems. We must hold the technology to a high standard of stewardship.

 

The future is not a choice between technology and nature but an obligation to make technology answerable to nature. By doing so, we can ensure that the immense intelligence we build is actively deployed to protect and preserve the natural world it depends upon.

Napa Vision 2050’s Monthly Column in Napa Valley Register