))
S2 #8 'How do we unravel the true climate-cost of AI and ChatGPT?', with Kasper Ludvigsen
Manage episode 366029025 series 3410813
Content provided by GoSquared. All podcast content including episodes, graphics, and podcast descriptions are uploaded and provided directly by GoSquared or their podcast platform partner. If you believe someone is using your copyrighted work without your permission, you can follow the process outlined here https://player.fm/legal.
It's not every day you get to sit down and pick the brains of a data scientist for The Danish National Police. Especially when that data scientist also happens to be an expert in the climate impact of AI. But today was one of those days!
Have you ever considered the carbon cost of AI? Every time you input a question to ChatGPT, how much energy do you think is required to run those processes? Then, zoom out a bit. Think about how many requests are made to ChatGPT every day across the world, how much energy is required to run that?!
Be prepared for this episode to take you down the rabbit hole of the true carbon cost of AI; machine language learning and processing.
In this episode Kasper takes James on a journey to investigate the questions no one is asking:
- What is the true energy cost of running AI programs like ChatGPT?
- How many litres of water(!) does Kasper calculate ChatGPT consumed in one month?
- How Chat4.0, OpenAI's successor to Chat3.5, could gobble over 5 times the energy required by Chat3.5!
- Why is it so difficult to get accurate calculations around AI's energy usage, and why are companies like OpenAI not more transparent?
- The importance of the right geographic location of AI's data centres.
- The correlation between drought-affected areas, and the installation of data centres.
About Kasper Groes Albin Ludvigsen
Kasper is a data scientist at The Danish National Police, with several years of experience in consulting and the financial sector. Kasper is also co-founder and board member at the Danish Data Science Community (DDSC).
Kasper cares about reducing the environmental impact of software in general and data science in particular. For instance by reducing the carbon footprint of training and serving machine learning models.
Further Resources:
Kasper on LinkedIn: https://www.linkedin.com/in/kaspergroesludvigsen/
Kasper on Medium: https://kaspergroesludvigsen.medium.com/chatgpts-electricity-consumption-pt-ii-225e7e43f22b
Kasper on Twitter: https://twitter.com/kaspergroes?lang=en
54 episodes
Share
