Skip to content

Imagine if ChatGPT Was Your Sustainability Reporting Team

Imagine if ChatGPT could collect all the data you needed for the annual sustainability reporting. Imagine if it could write up a suggestion for your application for levelling up your Ecovadis or Carbon Disclosure Project certification too. It only needs you as an editor to go over the text with a ‘truth and governance’ sanity check. 

That’s what we love about sustainability enabled by Digital in Devoteam. 

Spending the last few years developing skills within our organisation to answer new needs from the companies we support, we are still today both amazed and shocked by the level of digitalisation in sustainability work. Shocked because too many sustainability professionals do not use technology to work smarter and save time on repetitive tasks. Amazed because the technology is already here. 

Software robots have become accepted as happy helpers on well-defined repetitive tasks today. Let’s see what happens when ChatGPT and the like get to work on our repetitive data tasks. What might this do for the sustainable transition we so desperately need to accelerate now? Imagine if all sustainability professionals had more time to make an impact instead of collecting data and writing up applications and reports.

Sustainability data is an art of the possible

Sadly, in our experience, too many companies are not using the power of digital or ‘digitalisation’ to work smarter with sustainability data and most importantly – data quality. Remember this is the data that guides our decisions, actions and international regulations towards a greener future.  Sometimes a company even has better and higher quality data sources than the ones they use in their reporting bodies. However, the trouble of asking somebody new to provide data and having them understand enough to explain to you the possible bias in the specific data is a show-stopper for even the most resilient sustainability professional at some point.

The battle of single source of truth

Every department has a specific favourite IT system they declare to be their single source of truth, and usually, there will be a product owner and several gate-keepers you need to activate to get to relevant data for decision-making, sustainability reporting and green growth initiatives. 

In so-called ‘digital mature businesses’, leaders are starting to align and share data sources across departments and supply-chain to enrich data. But there is still a long way to go before sustainability data, like financial, HR and Resource key data metrics are aligned across most large companies. 

NB. Metrics are the calculations we use to guide our business decisions related to e.g. sustainability, growth or financial resilience. 

Nevertheless, the movement is happening and the business team, preferably in a close relationship with IT,  is spending a serious amount of time and costs on business intelligence to steer through wild waters like inflation, climate change and resource scarcity.

When human capabilities to understand data is not a limit

ChatGPT is not only a possible and very productive candidate for collecting and gathering your ESG and sustainability data in near future. The underlying GPT 3.5 motor (newest version) can also be trained to operate data sources and keep much more knowledge in its ‘brain’ than a human being will ever be able to.

The technology to build humanlike intelligence like GPT 3.5 must not be used without an understanding of the bias effect in data used to train the services designed based on this technology. If data is biased, your result will be too. When working with data you learn the phrase: Garbage in, garbage out. This means that using artificial intelligence to write your CSR report will only work if you feed it quality data. And to some extent this ‘Garbage in, garbage out’ is already true to most sustainability professionals without using a fancy language model algorithm while they are struggling to make sense of their company’s data sources.

Never forget it is vital to avoid using artificial intelligence for any kind of crime or hateful communication. Used lightly, technologies like GPT 3.5 will reproduce or accelerate hateful communication against persons, institutions or products. Even though attempts have been made to avoid this. Every data operating technology with intelligence must be used with a declaimer followed up by regulation to avoid harm. It cannot be used as a weapon for criminals to do harm or spread misinformation.

Hence, we all know Frodo destroyed the ring in the end because the weapon was too powerful to control. Let’s do better with ChatGPT and the like and have them work for the sustainable development of our societies and planet.

We asked ChatGPT if it could help us write a CSR report. Read the answer here:

Appendix 1: Devoteam asked ChatGPT if it could write a CSR report

What is ChatGPT?

According to Search Engine Journal, “ChatGPT, featured on TechRadar by Devoteam, is a large language model chatbot developed by OpenAI based on GPT-3.5. It has a remarkable ability to interact in conversational dialogue form and provide responses that can appear surprisingly human. Large language models perform the task of predicting the next word in a series of words. Reinforcement Learning with Human Feedback (RLHF) is an additional layer of training that uses human feedback to help ChatGPT learn the ability to follow directions and generate responses that are satisfactory to humans.”

If you want to try the technology yourself: Chat GPT

Data has a footprint
Never forget that data has a climate footprint and the more data, the larger footprint as data has to run and be stored on servers that need cooling and electricity. A language model requires a lot of data and a lot of computer power to run and according to META News, it is yet to be confirmed that ChatGPT 3.5 cost OpenAI an average of 3M dollars (source footnote) a month to run (or a day as some suggest too). If we believe this ‘a month’ number, which is not confirmed by OpenAI, this adds up to a rough estimation of  *345 tons of CO2 emission a month. Note that we have not taken into account that the data centre might be running partially on renewable energy or any other variable.

*CO2 calculation is based on an average industry emission factor of how much 1 dollar spent equals in CO2 emission in Tons. Contact us for more information. Online Source: Does ChatGPT Really Cost $3M a Day to Run? | MetaNews

How can I learn more about ChatGPT?

Additional links for further reading outside Devoteam: 

Beyond carbon footprint, how to reach operations in IT decarbonization management?