91Â鶹¹ú²ú¾«Æ·×ÔÅÄ

The fifth webinar in UNESCO's Business Council webinar series on artificial intelligence (AI) and ethics, held right after the conclusion of COP28, focused on the dual theme of making AI greener and leveraging AI for environmental sustainability. The discussion brought together experts from Telefónica and Microsoft, the co-chairs of the Business Council, as well as insights from Salesforce, which recently joined the Council. Participants included Boris Gamazaychikov, Senior Manager of Emissions Reduction at Salesforce, Richard Benjamins, Chief Responsible AI Officer at Telefónica, Maria Paula Duque, Sustainability Lead in Latin America at Microsoft, and Angel Melguizo, AI Senior Advisor at UNESCO. We discussed diverse strategies for reducing the environmental impact of AI technologies and utilizing AI to address climate change and biodiversity conservation. This blog post synthesizes its discussions, highlighting key insights and examples of AI's role in environmental sustainability.

Boris Gamazaychikov from Salesforce highlighted the company's efforts to address the environmental challenges presented by the computational demands of Large Language Models (LLMs). He emphasized the importance of developing smaller, domain-specific models, enhancing efficiency, and deploying these models on hardware in data centers powered by renewable energy to reduce the carbon footprint. Salesforce's commitment to sustainability is reflected in its collaborative effort involving the AI Research, Sustainability, and Office of Ethical and Humane Use teams, focusing on optimizing models for efficiency, utilizing advanced and efficient hardware, and prioritizing training in low-carbon data centers. This strategy is part of Salesforce's broader goal to align its technological advancements with its core values—trust, customer success, innovation, equality, and sustainability—and to reduce emissions in line with the 1.5-degree climate target. 

Richard Benjamins from Telefónica outlined the company's commitment to fighting climate change through AI and data analytics, illustrating its approach with initiatives aimed at enhancing air quality, optimizing urban low-emission zones, and evaluating the economic impacts of climate policies. Telefónica's dual strategy emphasizes using AI for green purposes, including predictive analytics for air pollution at the urban level by managing to combine mobility data from the network with open data in cities such as Sao Paulo, Brazil. Richard detailed the development of tools to measure and reduce the carbon footprint of AI algorithms, reflecting a commitment to the goal of achieving net zero emissions by 2040. He also highlighted the challenges of scaling such initiatives, underscoring the need for investment and collaboration among private and public sectors to realize the full potential of AI in supporting environmental sustainability. 

Maria Paula Duque from Microsoft highlighted the transformative role of AI in achieving the SDGs, with a focus on nature-based solutions and accelerating the journey towards net-zero emissions. By utilizing AI, Microsoft seeks to bridge gaps in environmental monitoring and management, thereby enhancing biodiversity conservation and climate change mitigation efforts. Maria Paula shared insights into a pioneering project in Colombia, Project Guacamaya, where AI analyzes sound and image data from the Amazon rainforest to enhance the efficiency of scientific research and the quality of carbon bonuses that will be emitted for Amazon protection. She underscored the potential of AI in climate action, estimating that AI could mitigate 5-10% of global emissions by 2030. She also emphasized the importance of AI in driving sustainability solutions, from predicting water consumption in agriculture to providing weather services that assist companies in adapting their operations to climate variability. 

Angel Melguizo from UNESCO reflected on the outcomes of COP28, emphasizing the critical role of AI in addressing environmental sustainability and the challenges posed by its energy consumption. He advocated for a conscientious approach to deploying AI, stressing the importance of assessing its direct and indirect environmental impacts, including energy usage, raw material consumption, and water usage for cooling data centers. At the same time, different AI-driven innovations on smarter mining and energy savings innovations in datacenters (AI greening itself), optimized air traffic and urban transportation (AI greening the economy) and biodiversity preservation (AI keeping green green) were described, Highlighting the need for transparency in reporting these impacts, Angel underscored UNESCO's recommendations for a balanced use of AI, advocating for cost-benefit analyses to determine when AI is truly necessary and urging for developing smaller, more efficient AI models.