As a recent climate BioAI quitter who is aware of (and slightly tempted by) the trends of using ChatGPT to make action figures of ourselves, I often find myself morally torn. How can I indulge in the cheap and comforting dopamine hits of AI image filters, whilst pretending not to worry about the costs that are deliberately hidden from plain sight?
Apparently companies hide this information because they don’t want to ‘spark our concern’, according to the Guardian🫠. That’s nice sweetie, but that’s called green-hushingto give people the illusion it costs nothing for our convenience. But it’s actually a lack of transparency, which means that we never really know what we are buying into, yet we continue to feel responsible.
No wonder we feel bad, because it’s actually a tad muggy. 😬
This black box 🗃️ of AI’s impact on the planet and the elements is called Digital Materiality and is notoriously hard to measure and therefore to optimise. It isn’t just about carbon, but water usage to cool the servers 💧, rare earth elements that are mined for it’s making 🪨… not to mention the new wave of modern day slavery of low paid workers in the global south whom label the global norths elixir; the data used in AI training. Technology should make you feel better and on top of the world, not helpless and blamed for burning it.
AI is a Mammoth
Just to paint a picture of it’s footprint - In the elements of energy, water, waste, raw materials, carbon emissions and impacted communities, AI is a mammoth when it comes to energy consumption in training (the people that create chatGPT) and usage (us making images). But whilst doing this, it also heats up the data centers so much, it requires extensive water to cool it down. This uses A LOT of water. Think of AI training like a big lawn that needs lots of water for cooling, only this time it isn’t replenishing the grass, but plunders through physical resources (finite) to create packaged digital information (infinite).
Each ChatGPT query needs nearly 10 times as much electricity as a typical Google search, which often comes from burning fossil fuels like coal and gas, which are primary drivers of climate change….Not what we signed up for.
As a society which are becoming increasingly reliant on technologies, however, we often don’t have a say in how they are made. What we buy into currently stands as our greatest source of individual power to advocate for peaceful, green options without the risk of being arrested . Either way, our hands seem tied. 🪢
Much like myself upon climate quitting, it seemed that the best of a companies efforts were to reduce, reduce and reduce in the hope that tentatively doing more of the same will magically decouple productivity from resource 🪄🎩 . We will keep on using it, so we need a better way to build it.
Hope in this context is a futile sport. Making different choices will.
We need an accessible green AI option which gives back to the planet, whilst summarising our 60 page PDFs and let’s face it, gives us life advice. 🔮
Paying It Forward with GreenPT
In the midst of my eco-anxiety-fuelled curiosity, a post on LinkedIn led me to discover a better way forward, starting with GreenPT. 🌍 and the brilliant sustainable powerhouse behind it, Leafcloud.
I spoke with Leafcloud in early April 2025 and they said it all started when a group of students in a small apartment realised that their laptops were keeping the room warm. That sparked a thought — if computers give off heat anyway, why not use it? What if, instead of cooling data centres just to throw that heat away, we put servers in places that actually need warmth? From that simple question came a bigger idea: design digital infrastructure that works with the environment, not against it. And just like that, Leafcloud was born — not just a tech company, but a reimagining of how we power the cities of the future. (This isn’t ad btw, just plain, straight up, basic bitch fan-girling).
Why?
a) It uses renewable energy to power it’s use (*)
b) It reuses this energy by capturing the heat from the data centers and repurposes it to heat community pools, showers, and buildings. 🏊🏼♀️ 🛀🏽
Taken from Leafcloud, the leaf site bar is reversed because this energy is reused later.
It is exactly the right school of thought and approach to resourceful building, and how to design for reuse and circularising heat back to communities that need it, whilst powering algorithms on the other side of the world.
It’s on the right track and paves the way for how to be truly efficient.
Critics may say, the training algorithms weren’t built using renewable energy (much like chat GPT), but we have to start somewhere and for this to happen, this takes a lot of funding to create from scratch. For me, this is a win.
Excited by its potential, I joined the waiting list and messaged the founder to share my enthusiasm. They invited me to be an early tester, which I took on wholeheartedly as an unpaid passion project — just for the green giggles. 💚 💻
🌿 GreenPT: How Was It?
Over the past few weeks, I’ve been testing the TWO options in GreenPT, which you can use and switch in between the same tab:
GreenL: a LLM (Large language model) similar to chat,
GreenR: a reasoning model that breaks down tasks into action plans and strategises.
Screenshots from GreenPT
In short: I loved it!! 💚
The GreenL is basically like ChatGPT but less wordy and gets to the point quicker and GreenR was the resolve and clarity you wished you had from chatGPT after about 10 commands of prompting.
🌍🎙️ There’s now a brilliant podcast about GreenPT, where the Founder, Robert Keus, discusses the topic of what makes a sustainable AI and ‘Is GreenPT the most sustainable AI chat?’ 🎙️ 🌍
🧠 🧠 Deep Dive: For those that love technical detail 🧠 🧠
🛠️ How I tested it:
Assessing job applications — GreenL gave me clear summaries of expectations of the job description and compared this with my CV to see if it was a good fit for my skills and values, and if the company sounded inclusive.
Navigating legal forms (NDAs) — both L and R worked well, but R helped me frame action points and flag any concerns in contracts.
Support for neurodiversity questions — especially around autism and ADHD, Green R stood out with useful, sensitive responses. Knowing things like time blindness, executive dysfunction, or breaking down tasks — Green R “got it” in a way ChatGPT hasn’t always managed.
How to pivot into a new field - with some strategic steps using R, and gave resolve much faster than ChatGPT usually does.
Moral dilemmas - e.g How to not invite someone to a party, or how to tell a friend hard truths, as well as others. I found it was thoughtful in it’s response and gave you options and what it would look like to the other person. Useful, especially on GreenR.
✅ What I Liked:
Green R offers direct, actionable responses — no fluff.
Responses use clear formatting (bold headers, steps, bullet points).
There’s a noticeable awareness of neurodivergent needs — especially around processing difficulties and stress reduction which require sharp nuance. I felt informed with an action plan that understood me.
Switching between Green L and Green R depending on task type is really helpful — it’s modular AI, which feels smarter.
🧩 What’s the Difference?
🛠️ Suggestions for Improvement:
Introduce a “like” button to give feedback and highlight useful responses.
Refine accuracy, especially on public information or named entities (this will improve as time goes on)
💭💭 Final Thoughts 💭💭
GreenPT isn’t just a spin-off of ChatGPT — it’s a thoughtful redesign that considers how people process information differently and how digital resources can be repurposed rather than wasted.
Would I use it again? Absolutely. It’s still in early testing, but the more people are aware of alternatives like this, the easier it becomes to shift our digital habits to more regenerative options — even if just some of the time.
The big question is: If our technology can take so much, can we also find ways for it to give something back?
With the right momentum, platforms like these - GreenPT and Leafcloud - could help us build a smarter, greener, and more reciprocal tech future through example - which is surely something to feel pretty good about, I would even say that it ironically gives me some hope.
Thank you for reading this! I’d love to know your thoughts. If this resonated, feel free to like, share, or forward it to someone who’s also trying to make sense of tech, brains, and the planet — one tab at a time. 🌍
Takeaway Goodies 👜:
🎬 WATCH: If you want to research more into this, there is a brilliant video by Harmer Visuals that interviews the Founder of Leafcloud, where one of my favourite bits was when he mentioned that “most software engineers don’t know how datacenters work”! 🎬
💻 🌱 READ: If you are interested in the sustainable software space, check and connect out the awesome community Green Software Foundation 🌱 💻
🎧 LISTEN: If you are more of a podcast person, listen to this great podcast by Green IT on a Cloud with Wilco Burggraaf, a seasoned software architect who takes us on a deep dive into the world of Green Coding—a transformative approach that prioritises energy efficiency and resource conservation in software development. 🎧
Dates with Divergence - Dr Camilla Pang is a reader-supported publication. To receive new posts and support my work, consider becoming a free or paid subscriber.
