The rise of artificial intelligence (AI) has significantly increased energy consumption and carbon emissions among major tech companies. A new study published in Frontiers highlights the environmental impact of different generative AI models, revealing that smaller models, trained with fewer parameters, can produce up to 50 times less pollution than larger models but often lack accuracy. The research analyzed 14 open-source models, demonstrating that while larger models achieve higher accuracy, they also result in substantially greater emissions. A key takeaway is that for simpler tasks, smaller models are more efficient and environmentally friendly. The study aims to guide users in selecting appropriate models based on task complexity, encouraging the use of less polluting options to minimize AI’s environmental footprint. Researchers are also developing a tool to facilitate this choice. The findings underscore the urgent need for tech companies to adopt sustainable practices as their greenhouse gas emissions continue to rise.
Major AI Models Produce Up to 50 Times More Carbon Emissions than Smaller Alternatives
