AI is Powerful—but is it Sustainable? Here's How Everyday Users Can Make It More Ethical
Artificial Intelligence has become a go-to productivity tool for many of us in social impact work. But here’s the catch: AI is not free—at least not to the planet.
Training and using AI models consumes vast amounts of energy and water and contributes to electronic waste at staggering rates.
Training one large AI model can consume as much electricity as 100 U.S. homes use in a year.
Each AI query can use up to 10x more energy than a Google search.
Training GPT-3 is estimated to have consumed 700,000 liters of fresh water, largely to cool the hardware running training computations.
As demand for AI capabilities grows, so does the need for land to host these data centers. This could displace natural spaces, agricultural areas and communities.
And the ongoing use phase (inference)—not the training—is where the majority of energy is burned, daily, billions of times over.
As a carbon measurement company, we at Impacti know that you can’t manage what you don’t measure. But before measurement comes mindset.
If you're a social impact organization or an individual committed to sustainability, consider this everyday ethical AI framework:
Use AI only when needed: If your task is simple, don’t fire up an energy-hungry model. Save it for complex questions where AI’s capabilities are essential. Use web tools like Grammarly for basic editing or Deepl for translation. Save AI for the complex stuff: making sense of unstructured data, modeling scenarios, or solving problems where human and machine insight together create real value.
Choose greener models: Opt for small language models like DistilBERT, TinyLLaMA, Mistral 7B, and Gemma or platforms that disclose and minimize their environmental footprint like Anthropic’s Claude, Google Cloud, and Hugging Face.
Prompt wisely: Learn to design efficient prompts to reduce unnecessary computation—get to your goal faster, with fewer iterations. For example, a single well-crafted prompt can save up to 5 times the energy of a trial-and-error approach—when you multiply that by billions of daily queries, the environmental savings become massive.
Balance cost vs. benefit: Ask yourself, Does the productivity or insight I gain justify the carbon and water costs of this AI use? For example, if you're using AI to optimize logistics and reduce vehicle fuel use, the resulting emissions savings can greatly outweigh the emissions from running that analysis. For social impact organizations, this kind of tradeoff—where AI enables tangible environmental or social gains—is where its ethical use truly shines.
Ultimately, this comes down to measurement—of emissions, of efficiency, and of impact. At Impacti, we’re working on ways to help you do just that.
How are you grappling with the ethics of AI use in your work? What barriers are you facing?
Let’s build a better, measurably sustainable AI future—together.
#SustainableAI #EthicalAI #DigitalSobriety #CarbonAccounting #ImpactTech #ClimateAction
