Resilient AI Challenge

💻 As AI systems grow more powerful, they also consume more energy and computing resources. This raises important questions about sustainability, accessibility and global inclusion. The Resilient AI Challenge asks a simple but critical question: Can we build AI models that deliver strong results while using significantly less energy?

🌍The Governments of France and India, along with UNESCO, launched the Resilient AI Challenge during the India AI Impact Summit. This is the first international Challenge on AI models Compression. Participants are encouraged to develop the most efficient compressed model balancing quality and energy savings. Teams may compete in one or more of the following categories: Audio-to-text (using Voxtral Realtime by Mistral AI), Text-to-text (using a model from Sarvam AI) or Image-to-text (using GEMMA 3n by Google), each category focusing on a distinct real-world use case and model type to explore different compression techniques.

⚡ The Challenge introduces shared benchmarks that measure both model performance and energy efficiency, helping to establish new standards for resilient AI. The Challenge is open to the research community, companies, start-ups, and innovators who aim to advance the field of model efficiency. By participating in this international competition focused on AI model efficiency, you contribute to shaping a more accessible and sustainable future for artificial intelligence.

🏆 The winner of each category will be the team that delivers the most efficient compressed version of the baseline model while meeting defined thresholds for both accuracy and energy performance.

👀 Get more details at: https://www.sustainableaicoalition.org/resilient-ai-challenge/