MLCommons Launches Candela for AI Risk Testing
MLCommons has introduced Candela, a service that helps developers and companies check AI systems for risks. It is offered as benchmarking-as-a-service, so the testing is provided online rather than built in-house. The tool is aimed at both the people who build AI and the people who deploy it in their products.
Quick glossaryMLCommons, benchmarking-as-a-service, deployers
- MLCommons
- A US nonprofit known for shared tests that measure how well and how fast AI systems perform.
- benchmarking-as-a-service
- A paid online service that tests and scores a system against standard measures, so no in-house setup is needed.
- deployers
- Organisations that put an AI system into real use, such as in apps or services that customers use.
Burning Potato score: 70%