AI is becoming prevalent in many work environments, with accompanying claims around the technology’s positive and negative impacts on the climate crisis. A recent article in Transform [1] (ISEP’s bimonthly publication) spoke with Golestan Sally Radwan, the Chief Digital Officer at UNEP, about the current AI landscape and the safeguards needed to make it a positive future tool. As a data based environmental business, SHIFT takes an interest in technologies connected to these themes, and has summarized the article below.
Unhelpful Uncertainties
When weighing up the benefits of using AI, users are hampered by the lack of knowledge and scientific consensus on the technology’s impact. Radwan states there is ‘no common scientific understanding’ to base discussions on, allowing claims to go unchecked and responses to be emotional.
There are claims it will optimise the sustainability of systems, with a study by the London School of Economics [2] predicting large (3.2-5.4 GtCO2e) GHG savings in major emission sectors such as food and power. While the same report predicts AI’s own emissions are projected to grow by 0.4-1.6 GtCO2e. These values are based on assumptions, and so the potential net benefit they represent is uncertain. Even if these numbers prove true, there is the social aspect to consider, many have concerns around AI’s impact, with greater worry appearing in more vulnerable demographics e.g. women [3].
Tasks Maketh the Tool
To mitigate the projected harm, Radwan reiterates the need for international governance of AI. After all, it is a technological tool, the parameters and uses of which are dictated by humans. If used unethically or in contradiction to a sustainable future, such as discovering new fossil fuel resources, AI’s potentially negative impacts become a chilling reality.
However, if governance frameworks such as the UN General Assembly’s proposed one (built along the lines of another Paris Agreement) come into effect, AI may become the hoped for helper in pursuit of a more sustainable future.
Current Considerations
AI technology has evolved so quickly, that there is still much to be learnt about its environmental and social impact. Knowledge that will hopefully inform regulations around use and development, with data transparency that can be verified by impartial third parties.
While that future is shaped, Radwan recommends searching out AI models that use reliable databases to answer queries, such as the new environmentGPT [4] that has been launched by the UN. She also mentions considering if a job can be done by software with a much lighter resource footprint: mentioning basic data analysis as one of these tasks.
Such data activities are intimately familiar to SHIFT Environment, where we rely on company-developed automated methods and knowledgeable assessors to ensure clients receive the most accurate and impartial assessments, while keeping the associated CO2 footprint as low as possible.
Next steps
AI still has a way to go before it can produce excellent environmental reporting. Until then, we have streamlined our environmental reporting and strategy development processes to ensure high quality reports with excellent audit trails. This gives you great peace of mind when presenting your performance to stakeholders. If this is something you’d like to do, then please feel free to get in touch: [email protected]
[1] https://www.isepglobal.org/engage/transform/features/golestan-sally-radwan-ai-a-known-unknown/
[2] https://www.nature.com/articles/s44168-025-00252-3
[3] Global views about AI’s impact on society in the next 20 years, by demographic group, 2021
[4] https://wesr.unep.org/environmentgpt/
Image by Luke Jones on Unsplash