All about technology and forces shaping the energy transition in shipping

‘Climate Positive’ Holiday Greetings!

December 22, 2023

Let’s dive right into the math and geeky details behind the emissions footprint of my AI-assisted ‘holiday greetings’ LinkedIn post!

1. Firstly, the post and its reach:

  • The size of the LinkedIn post, including the text and image is estimated at 142 KB. For ease of calculation, I’ve rounded it up to 150 KB.
  • As it is the festive season, I decided to work with the best case reach scenario:
    my best performing LinkedIn post ever had 43,631 impressions, 292 reactions, 26 comments and 7 reposts. This totals to 43,956 interactions. I’ve rounded this up to 44,000 to account for any reactions to the reposts that I might have missed.
  • For the purpose of this exercise, all reactions have been considered equal; i.e., they consume the same amount of energy and consequently produce the same amount of emissions.
  • ChatGPT roughly estimated that the energy required for loading a 150 KB social media post on a smartphone with average efficiency might be around 0.3-1.1 Wh (considering device energy, network energy, screen energy and idle energy). I’ll work with the average of these two values, 0,7 Wh per impression/interaction.
  • Loading the social media post on a computer with average efficiency would require between 1.25 and 6.3 Wh. I averaged these two values to arrive at 0.57 Wh.
  • Assuming 85% of the interactions are on a smartphone, and 15% on a computer, the weighted average energy consumption per interaction of the 150 KB post is 1.16 Wh.

2. Onto the ‘network’:

  • I have a global network, and I assume that the geographic distribution of where the reactions have come from is a simplistic extrapolation of the location of this network.
  • I asked ChatGPT to estimate the carbon intensity of 1 Wh in different countries and then calculated the weighted average carbon intensity per interaction.
  • Based on these two elements, I calculated the Weighted Average Carbon Intensity of one interaction from my LinkedIn network. This cames up to 0.18 gCO2e/Wh.
GeographyAverage Carbon Intensity per Wh
LinkedIn Network DistributionWeighted Carbon Intensity contribution
Total 100%0,18
Weighted Carbon Intensity of an interaction from my LinkedIn network
Calculated based on data provided by ChatGPT on grid intensity. Data years 2020 or 2022

Based on 1. and 2. above, the emissions from this post would be 1.16 Wh*0.18 gCO2e/Wh*44,000 = 9.13 kg CO2e

3. Thirdly, the energy consumed during the production of this post:

  • The work on this post took about 8 hours.
  • I used a MacBook Air for ~87% of the time, a recent iPhone for ~11% of the time and a Windows laptop (older than the other two devices and less efficient) for about ~2% of the time.
  • After failing to get DALL.E to remove the smokestack and smoke from the original graphic of the cargo ship, and not succeeding in doing it convincingly without installing heavy software programs that were not yet on my MacBook, I had to use Paint 3D on the Windows laptop to hide the smokestack and the smoke. I don’t see smoke in Shipping’s future 😉
  • I assume that the lower energy consumption of the iPhone balances out the higher energy consumption of the Windows laptop, and therefore ignore the difference in energy consumption of these devices.
  • 8 hours of work at 50Wh (energy consumption of a MacBookAir, moderate use) would produce 0.16 kg CO2e (using NL grid intensity from above table, even though I have an all-renewables electricity plan).

Total expected GHG footprint of the production + LinkedIn interactions of post:
9.13 + 0.16 = 9.3 kg of CO2e

This website is hosted by GreenGeeks and powered by renewable energy. But I do not yet have the functionality to estimate the impact of you reading this detailed analysis. To account for this, I decided to compensate for 10x the above-calculated amount.

4. Compensating for the emissions:

  • I purchased 100 kg of CO2 from Climeworks’ Orca project in Iceland, which captures CO2 directly from the air and stores it underground.  
  • This is only one example of the many sci-fi-like mitigation, removal and alternative fuel solutions that needs to be developed with care and deployed responsibly in order for our planet to endure and prosper in the years ahead.

I learnt a lot while writing this, and I hope it gives you some new insights.

Happy holidays!!

Cover image: DALL.E
Cover image compression: Optimizilla
Research: ChatGPT

Disclaimer: This GHG accounting exercise was just a fun way to explore working with AI tools that have been at the top of our minds in 2023. It does not follow prescribed standards for specific industries or guidance frameworks (like the GHG Protocol), but attempts to emulate the accounting logic for the activity at hand.

By using this website, you agree to comply with and be bound by these terms and conditions.