CarbonLayer explores potential for standardizing AI environmental impact reporting

CarbonLayer is developing the concept of an AI Impact Receipt to document the environmental effects of AI operations. The proposed receipt would contain data on workload, model identity, performance, and emissions, along with methodology and uncertainty. The core challenge is creating a common reporting interface without forcing providers to use identical calculation methods, as their systems and data vary. The goal is to enable clear exchange of what is known and unknown about an AI task's impact, ensuring transparency over false precision.