What AI Art Actually Costs
So, About Those AI Avatars…
Lately, every time I log onto social media, my feed’s full of cartoon “self-portraits.” You know the ones: a person asks AI to create an image of themselves, given everything AI knows about them. They’re clever, colorful, and can be wildly addictive. And yeah, I get the appeal—who isn’t curious to see themselves as AI sees them?
But while we’re having fun, there’s something missing from the conversation: the actual cost.
Let’s skip the guilt trip. If you’ve already dabbled in AI-generated images (I have too), you’re not a villain. But now’s a good time to ask: what happens behind the screen—and is there a better way to do this?
This Isn’t “The Cloud,” It’s a Warehouse
We love calling it “the cloud,” like our data lives in some gentle sky puff. But AI tools run on massive data centers—think giant warehouses packed with high-powered servers.
These machines:
Suck up huge amounts of electricity
Generate loads of heat
Require nonstop cooling (which often means millions of gallons of water)
That water doesn’t appear out of nowhere. It’s pulled from nearby rivers, aquifers, or municipal supplies—oftentimes in places already struggling with water scarcity. So every time a trend like “AI Avatars of You” goes viral? That digital playtime triggers a very real spike in demand on local resources.
A Picture is Worth More Energy Than Words
As someone who’s worked in digital art and design for years, I’ve got strong feelings about this part. Generating images with AI isn’t the same as asking a chatbot for dinner ideas. It’s way more intense—technically speaking and ethically, too.
Here’s the short version:
Text generation = relatively lightweight
Image generation = heavy-duty (lighting, detail, composition, color—all those pixels add up)
The more complex the ask, the more energy and cooling required. Multiply that by millions of people jumping on a trend? You’ve got a serious spike in power and water use. Again, not about shame—just scale.
The Human Layer: Artists Are Feeling It
Here’s the part that hits close to home. Many of these tools are trained on artwork made by real people—without their permission, payment, or credit. That’s not “innovation.” That’s appropriation.
And when fast, free AI images flood our feeds, it makes it even harder for working artists and designers to get noticed, let alone paid. I’ve seen gigs dry up, contracts dissolve, and clients swap creativity for convenience.
Can AI help brainstorm or mock up an idea? Sure. But if it’s replacing actual creative labor, we’re crossing a line. And that line matters.
Smarter, Lighter, Kinder: Tips for Using AI Responsibly
Like I said—I use AI, too. But I try to be mindful about how and when. If you want to reduce the impact on the planet (and people), here’s how to start:
Make it count: Don’t generate ten versions just because you can. One solid image? Usually enough.
Stick to text when possible: Need help writing, brainstorming, or planning? Text-based AI uses way less juice than image tools.
Pay humans when you can: Commission art, hire designers, buy prints.
Sit a trend out: Not every viral wave is worth riding. You're not missing anything essential by skipping it.
Choose better platforms: Support companies that invest in renewable energy, ethical training practices, and transparency. User demand really does push change. (Check out some of my platform suggestions below.*)
It’s Not About Quitting, It’s About Choosing
AI isn’t evil. It’s not going anywhere, either. But like everything else, we get to choose how we engage.
If we already think about water use, energy, and labor when we’re outside, why wouldn’t we do the same with our digital lives?
That avatar might look like magic. But it runs on real electricity, pulls real water, and has ripple effects in real people’s livelihoods.
Use the tools. Just use them better.
*Use Better Platforms When You Can
We don’t have to ditch AI entirely—but we can choose tools and platforms that are trying to lessen the load. Some are making real progress on renewable energy and artist rights. Others? Still in the slow lane. Here’s a quick cheat sheet:
AI isn’t going away, but we can nudge it in a better direction. These platforms are making real moves toward sustainability and artist ethics—but none of them are flawless.
Adobe Firefly
Best For: Images with strong usage rights
Why It’s Better: Trained only on licensed/public domain art—no sneaky scraping
Caveat: Still part of a massive tech company; not all Adobe tools are this buttoned-up
Sustainability: Hit 100% renewable electricity for global ops in 2025
www.adobe.com/sensei/generative-ai/firefly.html
Runway ML
Best For: AI video, concepting, and creative editing
Why It’s Better: Transparent training data and creator comp (in some cases)
Caveat: Creator payment is limited, and platform scale means higher resource use
Sustainability: Hosted on AWS (Amazon Web Service), which hit 90% renewables in 2025
Microsoft Azure / OpenAI
Best For: Text-based tools (like ChatGPT), and image generation
Why It’s Better: Microsoft hit its 100% renewable energy goal in 2025
Caveat: Still hosts a ton of energy-intensive workloads (hello, viral image trends)
Sustainability: Carbon-negative goal by 2030; decent transparency
www.microsoft.com/sustainability
Google Cloud AI
Best For: Building and hosting custom AI models
Why It’s Better: Aggressive push for 24/7 carbon-free energy by 2030
Caveat: Google has faced criticism on water use and data privacy
Bonus: Publicly shares energy and water usage stats
cloud.google.com/sustainability
Want to Skip AI Art Altogether?
These platforms focus on real humans, not scraped data or algorithms.
Stocksy – artist-owned, fully human-made photo co-op
Death to Stock – real-life visuals from real-life photographers
Noun Project – icons and illustrations, all credited to individual creators
You don’t need to be perfect. Just thoughtful. Choose tools that align with your values, ask a few more questions, and maybe skip the next AI avatar trend that shows up in your feed.
Want more tips like this delivered in the least preachy way possible? Sign up at www.ExploreFieldTrip.com
