Policy Connect and the All-Party Parliamentary Group on Data Analytics recently partnered with Carbon Connect to host a roundtable discussion on the dual relationship between artificial intelligence and climate change.

Chaired by Daniel Zeichner MP, the roundtable considered both the positive ways in which artificial intelligence (AI) can be used to combat the climate crisis as well as exploring the environmental costs of Large Language Models (LLMs).

Participants noted that focus should shift from discussions of whether the benefits of AI outweigh the costs, or vice versa, but instead to developing AI models sustainably and responsibly. The session also discussed the role innovators, company directors, and policy makers can play in incentivising greener AI. 

The discussion initially focused on both the cost AI imposes on the environment and the role it can play in combatting the climate crisis. The insights delivered by panellists led to a conversation on the challenges in striving for sustainable AI and the environment required to foster sustainable and responsible AI.

The roundtable featured contributions from:

• Dr Gabrielle Samuel, Kings College London
• Dr Emily Clough, Ada Lovelace Institute

The key takeaways included that:

• Data-driven technologies can provide important climate solutions and play an important role in decarbonising sectors in which this may prove difficult.
• AI imposes a cost on the environment, and it is anticipated to get worse as AI is integrated further into everyday technologies. Caution is necessary however when discussing the environmental impact of AI.
• There are several challenges in striving for sustainable AI – mass competition; environmental costs throughout the supply chain; and lack of incentives and awareness.
• Innovators, company directors, and policy makers all have a role in creating an environment that fosters sustainable and responsible AI.

For more information, please contact alainah.amer [at] policyconnect.org.uk (alainah[dot]amer[at]policyconnect[dot]org[dot]uk)