Our Growing Reliance on Big Data

I hate squandering energy even as I rely on big data centers every day.

(Listen to the radio version here.)

Wouldn’t it be wonderful if this really was “the cloud”?

Russ and I do our level best to minimize our consumption of natural resources. It’s easy to see exactly how much gas we use in my Kia Niro hybrid, how much electricity and water we use in our house, and how much electricity our solar panels produce. But the energy that we use via big data centers is impossible to see on our own energy bills or any other way, even as Big Data mushrooms into a larger and larger portion of virtually everyone’s energy consumption. With the exception of DVDs, CDs, vinyl, and other hard copies, every video, movie, or program we watch on our computer, phone, or tablet and every bit of music and radio we listen to via streaming comes to us through big data centers. Every time we use the internet for anything, or many of those “smart” home appliances, we’re using Big Data. The so-called cloud where we back up data and save files, and where every one of my photos on Flickr resides, is hardly up in the sky—a tech “cloud” is just a pretty word for a data center.

It’s endlessly frustrating to me that there is no way to know how much energy and water I’m using now. I resent that so many apps I use are raising prices because they’re adding AI bells and whistles that I don’t want to use. And it drives me nuts that more and more social media users, including well-meaning people and organizations who support a clean environment, are now using very energy-costly AI-generated images and avatars—they’re suddenly inescapable.

Isn’t reality pretty enough?

I’m constantly pleading with people to stop using generative AI to create images for uses in which real photographs would work just as well, and to turn off AI on their search engines when possible. But meanwhile, whenever I upload a blog post to Substack or a radio/podcast recording to my website, I’m putting them on a big data server, and every time someone reads or listens to my work, they’re using big data energy, too. Data centers are ginormous users of electrical power and water—two things we simply can’t afford to squander.

Yet, ironically for someone who hates how energy-intensive it is, I use big data almost every day, both when I’m outside birding and when I’m home again, processing my photos.

In 2010, when the extraordinary bird photographer Shawn Carey and I joined forces (well, we shared the cost of a small airplane and a boat) to document the BP oil spill in the Gulf of Mexico, he suggested that I start using Adobe Lightroom to tag, organize, and process my photos.

The only photo I took of Shawn Carey in the Gulf. (We were both too busy documenting how bad it was to focus on each other.)

Yep. It was bad. And this poor Black-crowned Night-Heron didn’t figure into BP’s numbers of oiled birds because they only counted birds found dead or retrieved for rehab. Our boat captain was trying to legally retrieve this one right up until the moment we got close and it opened its wings and fluttered a few feet before falling into the water and struggling to wade to shore. The moment it opened its sticky wings, it was doomed; that met BP’s definition of “flighted,” and the captain would have lost his license if he were caught capturing a “flighted” bird. Anyway, thanks to Lightroom, I can quickly and easily find this picture by clicking “Black-crowned Night-Heron,” “Gulf of Mexico,” “BP oil spill,” “Louisiana,” “oiled bird,” or several other tags.

Lightroom was a game changer for me, and I’ve been using it ever since. This was back when you bought software rather than renting it, when Lightroom cost less to purchase outright than the monthly rental fee is now. (I use another Adobe product, Audition, to produce my radio show/podcast and to edit my natural recordings. I used to use Cool Edit, a shareware program that was easy for a non-tech person like me to learn, but Syntrillium sold it to Adobe in 2003. Now Lightroom and Audition are elements of Adobe’s “Creative Cloud.” None of my sound editing involves the cloud until I upload recordings.)

I’m a birder, not a photographer. I want my camera to work like binoculars only with a memory. No way do I want to fool with camera settings when I’m looking at birds, so I point and shoot in almost every situation. The only setting I ever mess with in the field is exposure—I overexpose when a bird is backlit…

I set my camera’s exposure at +1 2/3 stops to get this Barrow’s Goldeneye swimming in Duluth.

To get the detail on this backlit Northern Hawk Owl in the Sax-Zim Bog, I had to set my camera to overexpose by 2 1/3 stops.

… and underexpose in dark situations.

Cameras usually overcompensate in low light, so I underexposed this photo of a Cuban Nightjar by 3 stops. It was illuminated by the guide’s flashlight.

I set my camera to underexpose this photo of a Western Screech-Owl in Southeast Arizona by 2 stops. This bird was further away than the Cuban Nightjar, so the guide’s flashlight appears less intense.

With my DSLR cameras, I used to have to guess about the exposure. Now that I have a mirrorless camera, I can see for myself whether I need to tweak it.

This Andean Cock-of-the-Rock was very close, but the bright background made him look a little dark so I upped the exposure by 2/3 stop.

I didn’t tweak the exposure on my camera at all to photograph this Great Gray Owl in the Bog.

Except for exposure, I let my camera choose the best aperture, shutter speed, and ISO within the minimums and maximums I keep pretty constant.

Photographing the birds in the first place doesn’t involve any data centers. I upload them to my computer using “Lightroom Classic,” which doesn’t involve storing anything on the cloud. And unlike Photoshop, Lightroom doesn’t have many AI processes, but one recent addition, “Denoise,” has become part of my photo-editing routine.

To photograph birds, it’s normally best to use the fastest shutter speed and smallest aperture possible. Unfortunately, that forces the camera to use a high ISO setting, making pictures appear noticeably grainy, a problem exacerbated the more you zoom in. Some of my pictures that were otherwise fine were too grainy be used except as documentation. That’s why I now use Denoise.

This photo of a Crimson-rumped Toucanet was taken in early morning in Ecuador. I cropped and adjusted the lighting, then made a copy on which I used Denoise. If you click on them, you can see how much grainier the upper one, without Denoise, is compared to the lower one.

I photographed this Blue-billed White-Tern in 2024, before the AOS split the White Tern into three different species and created new names for each. The two photos are identical except that I used Denoise on the lower one. The difference is subtle but very noticeable to my eyes.

Because I have to crop pictures of distant birds so much, some otherwise useful photos are too small for PowerPoint presentations, my blog, or printing. If I need one for those purposes, I can use another AI process called “Upscale” from an app I bought from Topaz Labs. I hardly ever use it—I don’t usually mind small photos, and this only works well when the photo is very sharp and clear…

I shot this Red-footed Booby in the Dry Tortugas in excellent light, but from a distance, so the original photo is cropped to be only 988 pixels. After using Topaz AI to upscale it, it’s now 3952 pixels—plenty big enough for all kinds of uses.

…or when the subject is an important one for me, and I’ll be using that photo for a lot of purposes.

When cropped, the photo of this distant Andean Condor was 1085 pixels. I upscaled it to 4275 pixels.

These two AI processes involve fairly straightforward machine learning, meaning their energy consumption and the water essential for cooling the data centers aren’t nearly as intensive as generative AI—getting AI to create written or illustrated work or do a complicated search. But unless we start pressuring the companies to reveal exactly how much energy and water each process uses, how can we make responsible choices about whether and how often we use even relatively lower-use processes?

Right now, tech companies are as secretive about their data centers and energy usage as Donald Trump is about his friendship with Jeffrey Epstein—in both cases, the purpose is to hide information from Americans who have a right to know. At this very moment, Google wants to build a huge data center in Hermantown, Minnesota, right outside Duluth, but they kept their company’s identity secret for many months and forced the Hermantown elected city officials to sign NDAs about the project, even though those officials would be voting on a project that would affect virtually every person in the area.

Democracy dies in darkness. It’s impossible for us to make wise choices that will affect our own well-being and the world we’re leaving our children and grandchildren unless we have transparency. Tech companies should be required to reveal how much energy and water each of their products uses, and we consumers should be given clear and easy “opt-outs” for the AI elements of every app and search engine we use, and we shouldn’t have to pay for things we don’t use.

I’m not holding my breath about any of this changing. The many ways Project 2025 has already been implemented prove that the billionaires driving our government absolutely hate transparency. In his first term, Donald Trump eviscerated the popular Energy Star program which requires manufacturers to place a sticker on every new appliance showing exactly how much energy it uses so shoppers can make easy comparisons. Since 1992, Energy Star has saved an estimated 5 trillion kilowatt-hours of electricity, more than $500 billion in energy costs, and prevented 4 billion metric tons of greenhouse gas emissions. That savings for us and the environment was money insatiable billionaires wanted, and they’re the ones in the driver’s seat right now. There’s no way they’ll allow the politicians they own to expand the Energy Star program so we can see, and maybe even figure out how to conserve, the energy we’re using at their remote data centers.

Tomorrow I’ll discuss a very ironic way that a lot of us, including me, are squandering energy at big data centers, not while we’re sitting at our computers, but while we’re outdoors enjoying nature.

My ears can no longer detect the very high frequencies of many birds’ songs. I must use an AI-intensive app to know if one of them, like this LeConte’s Sparrow, is near.