ChatGPT

In response to the prompt "What are the environmental impacts of generative AI? Present in easy to understand language", MS Copilot in Edge, which integrates GPT-4o (Khan, 2024), generated the following response:

Generative AI, like the kind used to create text, images, and videos, has some significant environmental impacts. Here are the main points in simple terms:

1. Energy Consumption: Generative AI requires a lot of computing power, which means it uses a lot of electricity. This is especially true for training large AI models, which can consume as much energy as hundreds of homes¹.
2. Carbon Emissions: The electricity used by AI models often comes from fossil fuels, which release carbon dioxide (CO2) into the atmosphere. This contributes to climate change. For example, training a big AI model can produce as much CO2 as several round-trip flights between major cities².
3. Water Usage: AI data centers need to be kept cool, and this often requires large amounts of water. In some places, AI data centers use a significant portion of the local water supply¹.
4. Resource Depletion: Building and maintaining the hardware for AI, like servers and data centers, requires a lot of natural resources, including rare minerals. This can lead to environmental degradation and resource scarcity³.

¹ Generative AI’s environmental costs are soaring — and mostly secret

² Environmental Impact of Generative AI | Stats & Facts for 2025

³ Considering the Environmental Impacts of Generative AI to Spark Responsible Development – MIT Climate & Sustainability Consortium (Microsoft, 2024)

With respect to water usage, here's a way to think about how much one inquiry requires: In response to the prompt "How many bottles of water does one gen AI inquiry require?", MS Copilot in Edge generated the following response:

The water consumption for a single generative AI inquiry can be surprisingly high. For instance, generating a 100-word response from an AI model like ChatGPT can use approximately half a liter (500 ml)¹ of water. This water is primarily used for cooling the data centers where the AI servers are housed.

To put it in perspective, a simple conversation with an AI model consisting of 20 to 50 questions can consume about the same amount of water as a 500 ml bottle². This highlights the significant environmental impact of AI technologies, especially in terms of water usage.

¹ How Much Water Does AI Use? Too Much - Techopedia
² How AI Consumes Water: The unspoken environmental footprint | Deepgram (Microsoft, 2024)

Image retrieved from Pixabay

References

Khan, I. (2024, May 20). Microsoft’s Copilot embraces the power of OpenAI's New GPT-4o. CNET. https://www.cnet.com/tech/services-and-software/microsoft-copilot-embraces-the-power-of-openais-new-gpt-4-o/\

Microsoft. (2024). Copilot (Dec 18 version) [Large language model]. https://copilot.microsoft.com/