Have a burning desire to know something? Using less energy than it takes to brew a teaspoon of hot coffee, AI can deliver a detailed answer.

Just last week, OpenAI launched GPT-5, their most advanced artificial intelligence (AI) model yet, promising access to “Ph.D.-level intelligence” for everyone. Meanwhile, you’re trying to get up to speed on a complex issue like recycling post-consumer-waste plastic, but every article you find is either too technical or oversimplified. Here’s the paradox of our time: Do you take advantage of the newest technology to help you make well-informed decisions about environmental challenges or do you avoid AI but take longer to gather the knowledge you need to reach a deeper understanding of our ever-increasing environmental crises?

Environmentalists have legitimate concerns about AI’s energy consumption and footprint on our landscape. Data centers that consume enormous amounts of electricity and water for cooling are being built to develop even smarter AI systems. Don’t let this massive investment go to waste; learn how to use these new knowledge tools wisely so the benefits of your AI sessions can offset their environmental impacts.

What is AI and how can you use it?

Artificial intelligence “chatbots” are computer programs that can converse with you about any topic. Think of them as astoundingly well-read, but sometimes overeager, research assistants that have absorbed millions of books, articles, pictures, videos and websites, then learned to explain complex topics in whatever style you prefer. Simply type into a form on a website or speak to an app on your phone (“prompting” the AI). You’ll get a detailed, customized answer.

Popular AI tools are free to try, and they work a lot like texting or talking to a colleague. OpenAI’s ChatGPT (chatgpt.com), Anthropic’s Claude (claude.ai) and Google’s Gemini (gemini.google.com) are three options, each offering a simple signup. You can ask them anything from “Explain how solar panels work” to “Is drinking oat milk or almond milk better for the environment?” and receive responses tailored to your exact needs. The key is asking good questions and — crucially — verifying the answers, since these tools can sometimes make up incorrect information with complete confidence.

Understand what AI is really good at

You won’t be able to find a person in the world who is better read than an AI model. They have literally “read” every book ever published and have stored every bit of information they have ever encountered. Different models have different “personalities.” Most of the ones provided for public use are trained to do their best to try to understand what you want them to do and respond in a helpful and respectful way. They are especially good at explaining both sides of a contentious issue and considering every angle of a problem. Most AI models can be incredibly fair and balanced when discussing any sensitive topic.

Choose the latest AI models for better accuracy

“Hallucination,” occasionally making up stuff that is simply not true but seems like it should be true, can be a problem with AI. New models tend to hallucinate less frequently than older models, so it’s better to use a newer model. GPT-5, launched on Aug. 7, reasons better and hallucinates less than GPT-4, according to OpenAI, the company that created it. However, even the most advanced AI model will still make mistakes, so you can’t completely trust them without checking their answers. They are great study buddies, but just like study buddies in high school or college, they aren’t perfect. Your critical thinking skills are important to apply when you’re using AI.

Ask for research and references

Very old AI models (from way back in 2024) are not able to do research, so their answers to your questions are based on their vague recollection of everything they have digested during their training runs. It is interesting to chat with these older models because they provide a very good consensus view of any topic. But when you ask them specifics, they’ll often make up plausible-sounding answers that are wrong. New models (released in 2025) can do research and what’s called “reasoning.” If you tell them to do research and show their work, they can search the internet and find references that help ground their answers. You’ll still have to check their references, though.

When deciding how to gain knowledge, consider your alternatives. Searching the web, driving to a library or attending conferences — all have environmental impacts. If AI helps you learn faster and implement more effective sustainability practices, the environmental benefits may far outweigh the energy costs of the queries that enable them.

Peggy Siegle and Fred Horch are principals of Sustainable Practice. To receive expert action guides to help your household and organizations become superbly sustainable, visit SustainablePractice.Life and subscribe for free to One Step This Week, or go to suspra.com to purchase our book, “Sustainable Practices: Your Handbook for Effective Action.”