To me its clear, we have an urgent need to address the environmental impact of LLMs!

Recently, I attended the Google Cloud Nordic conference, where Ted Schönbeck shared some eye-opening insights into the environmental footprint of LLMs. He mentioned that ChatGPT consumes roughly 40x more computational power to generate an answer compared to Google Search. This finding aligns with a Forbes article that highlights the unsustainable energy consumption of these models, far exceeding the current supply of renewable energy sources.

I also came across an article written by Prof Wim Canderbauwhede this Christmas that suggests ChatGPT’s energy consumption could be in the range of 50x-90x (most likely 60x) compared to a standard Google search query.

I am en enthusiast and have been using many LLM’s extensively – as you might have seen on LinkedIn and my website. Lately I have reconsidered how I use LLMs and encourage others to do the same. Don’t get me wrong – we have transformational abilities in our hands – lets just do it the right way.

I believe that companies should prioritize sustainable practices to minimize their environmental impact. It’s one of that advices i bring with me today.

So what to do? That’s for my next article. 🙂

Links to articles and session:
“As ChatGPT And Other AI Tools Increase Energy Demand, Here’s What Investors Need To Know“ https://www.forbes.com/sites/petercohan/2023/11/09/equinix-and-vertiv-stock-prices-could-rise-on-generative-ais-energy-use/

“Emissions from ChatGPT are much higher than from conventional search“ https://limited.systems/articles/google-search-vs-chatgpt-emissions/

Session “Turning Generative AI into Business Value”: https://www.youtube.com/watch?v=Dr49BGY66Mg

Picture: me and puppy Bosse to get your attention. 🙂