In a recent article published in Nature, researchers explore the complex relationship between artificial intelligence (AI) and the global efforts to minimize climate change. As AI continues to reshape our world, understanding its environmental impact becomes crucial for policymakers, businesses, and researchers in different disciplines.

The Direct Impact of AI

In terms of total global greenhouse-gas emissions, the authors calculate that AI today is responsible for:

• Around 0.01%, on the basis of IEA assessments, showing that data centres and transmission networks together account for about 0.6%.
• This is less than cryptocurrency mining and conventional data centers combined.

However, the rapid growth of AI raises concerns about future energy consumption. With compute capacity for advanced language models increasing tenfold each year, and AI service demand projected to rise 30-40% annually, the energy needs of AI could skyrocket.

The Indirect Effects: A Double-Edged Sword

The indirect impacts of AI on climate change are more complex and potentially far-reaching:

Positive Impacts:

• Reducing emissions in energy and transport sectors.
• Optimizing supply chains.
• Accelerating development of new materials for renewable energy.

Negative Impacts:

• Increased electricity use from AI-embedded applications.
• Potentially cheaper oil and gas exploration, leading to increased production.
• Possible socio-economic disruptions affecting emissions indirectly.

The Need for AI-Driven Emissions Scenarios

To navigate this complexity, the researchers call for the development of policy-relevant scenarios that explore how AI might affect emissions. These scenarios should:

1. Integrate with existing climate models.
2. Consider both direct and indirect impacts.
3. Be regularly updated to keep pace with AI advancements.
4. Address data availability challenges in the fast-moving AI industry.

A Call for Collaboration

The article emphasizes the need for a multidisciplinary approach:

• Establishing an international consortium of experts from various fields.
• Creating standards for measuring and reporting AI-related data.
• Securing financial support from diverse sources to maintain ongoing research.

Conclusion

As we race towards net-zero emissions, AI emerges as both a potential accelerator and a possible roadblock. By developing comprehensive AI-driven emissions scenarios, we can better understand and harness the power of AI in our fight against climate change.