Searching, Prompting and the Planet: How might we use AI more mindfully?

Another week down and I’ve finally found a moment to pause and reflect. In the middle of running Microsoft 365 Copilot learning sessions recently, one question has kept popping up. And rightly so. Not “How do I get my week summarised to me in the style of a rap?” or “How do I actually get Copilot in PowerPoint to apply my brand guidelines?” (although that last one comes up more than I care to admit). Luckily, this time, it is something far more important.

How environmentally friendly is it to use? Should I be using Google instead?

It is a fair question and one that is always there in the background as I experiment and find the best ways of working with AI day to day. Should I be searching, prompting and what are the implications to the planet? Interestingly, some research suggests there is a generational shift in our search habits skewing us ever increasingly towards AI for search.

Younger users are leading the shift in starting-point behaviour, with 28% of Gen Z beginning their searches on ChatGPT (Adobe, 2025)

Gone may be the days of instinctively “googling it” when we want to find or understand something. With AI tools like Copilot and ChatGPT in the mix, we now have a choice, and those choices carry different environmental costs.

Does prompting with AI cost more to run than search?

Using Google for a simple search uses surprisingly little energy. You ask, it checks its vast index, and it sends back the answer in milliseconds. It is highly optimised and therefore highly efficient.

AI works differently. Each time you prompt, it generates a new answer for you there and then. It is like making a meal from scratch when there is a pre-prepared one ready and waiting on the side. This requires far more computational energy and rapidly adds up. One study suggests that “a generative AI system might use around 33 times more energy than running task-specific software” (Kelion, 2025). With the growing user base and volume of daily prompts, that difference could become significant.

But it's not that clear cut

There is some nuance here. Depending on the task, there is a case for which tool to pull out of the box. If it is a simple question, like identifying the actor in a certain film, search is probably a greener choice. However, if you are looking to summarise a report, translate a document or ideate for a workshop, the use of AI could save you and your peers hours, potentially avoiding multiple searches, clicks and time spent churning through several documents.

The recent challenge is that search engines themselves are getting more AI-driven. My recent Google searches have sometimes served me an AI-generated summary at the top of the results page. This is highly convenient, but it starts to blur the environmental line and the argument for using search or prompt.

So what might be the solution?

Some companies are showing their working, with firms calculating the cost of use. These figures do not make for light reading.

Google's carbon emissions have soared by 51% since 2019 as artificial intelligence hampers the tech company's efforts to go green (Hern, 2025)

Google has struggled to curb its emissions and is seeing its energy consumption increase each year as the demand for AI grows.

In my work and observations with innovators through to sceptics, encouraging the use of AI tools in the day-to-day has largely hinged on finding high-value use cases that make work more enjoyable, efficient or of greater quality. In doing so, it sparks enthusiasm to find more, and using AI becomes second nature.

The one area I am always encouraging is mindful use and critical thinking, ensuring outputs are reviewed and challenged appropriately. This is a skill we cannot afford to lose. I do wonder if we need something more visible to curb our enthusiasm to use AI as the default.

Would making the environmental impact more visible before prompting encourage greater thought into whether AI is the right tool for the job? Could our own computational power be up to the task, or could we take the time to use different sources to get to a great result? I often find the process of crafting an excellent prompt provides greater clarity on the ask and whether AI is likely to handle it well or whether I would be better doing it solo.

For now, how do I decide what to use and when?

Here's the guidance that I try to apply to my daily life.

Use search when:

• You're looking for a quick fact

• You're browsing options and ideas and have time to scan

• You're doing something straightforward and low effort

Use AI when:

• You're looking to synthesise information from many sources

• You're tailoring some output in a particular style or context

• You're going to save hours of extra work and other energy use

And no matter what you choose, you can reduce your footprint further by:

• Being specific in your prompts so you get to the desired result quickly, with the least amount of processing

• Choosing efficient tools, for example on-device AI, which can use less power than cloud-based models

• Batching your requests so you get what you need in as close to one go as possible

Why does it matter?

In the constant pursuit of instant gratification and never being more than a metre away from technology, it is easy to think our digital actions have no consequences. But behind the scenes, vast cooling systems are running 24/7, using millions of litres of water a year, and energy consumption is growing.

We do not need to be perfect, but we can be more mindful. A little awareness and self-challenge can help us work smarter, use AI responsibly, and work a little greener.

If there are ways you are encouraging mindful use that you would like to share, please let me know below.

---

References

Adobe. (2025) ChatGPT as a search engine: Benefits, drawbacks, and how it compares. Available at: https://www.adobe.com/express/learn/blog/chatgpt-as-a-search-engine (Accessed: 15 August 2025)

Hern, A. (2025) ‘Google’s emissions rise 48% in five years due to AI electricity demand’, The Guardian, 27 June. Available at: https://www.theguardian.com/technology/2025/jun/27/google-emissions-ai-electricity-demand-derail-efforts-green (Accessed: 15 August 2025).

Kelion, L. (2025) ‘ChatGPT: Why we should all be more mindful about using AI’, BBC News, 14 August. Available at: https://www.bbc.co.uk/news/articles/cj5ll89dy2mo (Accessed: 15 August 2025)

Credits

Author: Rob Anderton

Editorial: Rob Anderton / OpenAI. (2025). ChatGPT 5 [Large language model]. https://chat.openai.com/chat

Images: Midjourney (2025). V7 [Image generation model] https://www.midjourney.com/ / OpenAI. (2025). ChatGPT 5 [Large language model]. https://chat.openai.com/chat