DeepMind exec: AI assesses climate issues, falls short of full solution

Amid efforts by climate scientists and advocates to address environmental challenges, Google DeepMind Climate Action Lead Sims Witherspoon sees potential in artificial intelligence (AI), emphasizing the importance of framing the solution through thoughtful questioning.

At the Wired Impact Conference in London, Google DeepMind Climate Action Lead Sims Witherspoon said she sees climate change as a scientific and technological challenge, expressing optimism in addressing it through artificial intelligence. Earlier this year, Google merged its Brain and DeepMind AI teams under a single banner called Google DeepMind.

Witherspoon suggested a strategy dubbed the “Understand, Optimize, Accelerate” framework, outlining three steps for tackling climate change with AI, which involve engaging with those affected, assessing AI’s applicability, and deploying a solution for impactful change.

Examining the path to deployment, Witherspoon observed that certain options become less viable due to existing regulatory conditions, infrastructure constraints, or other limitations and dependencies such as restricted data availability or suitable partners.

Witherspoon stressed the importance of a collaborative approach, highlighting that while individual expertise is valuable, cooperation is crucial and necessitates the combined contributions of academics, regulatory bodies, corporations, non-governmental organizations (NGOs), and impacted communities.

Witherspoon said that, in collaboration with the U.K.’s National Weather Service Meteorological Office in 2021, Google DeepMind leveraged their comprehensive radar data to analyze rainfall in the U.K. using AI. The data was input into Google’s Deep Generative Model of Rain (DGMR) generative AI model.

Witherspoon stated,

“We conducted a qualitative assessment involving 50 meteorological experts at the U.K. Met Office, and over 90% of them favored our methods—ranking them as their top choice over traditional methods,”

Related: Google DeepMind AI predicts 2 million novel chemical materials for real-world tech

She emphasized that the source code data and verification methods are openly accessible. Despite recognizing AI’s potential in addressing climate change, Witherspoon also warned that this emerging technology is not a cure-all.

Sims Witherspoon said AI is not a universal solution for climate challenges. She underscored the importance of deploying AI responsibly, acknowledging its environmental impact due to energy-intensive processes until the grid operates on carbon-free energy.

In May, Boston University’s Kate Saenko warned about the environmental impact of AI models like GPT-3. The 175 billion parameter model consumed energy equivalent to 123 cars for a year, generating 552 tons of CO2, even before its public release.

Magazine: Real AI & crypto use cases, No. 4: Fight AI fakes with blockchain