Skip to main content
In this interview, we ask Barry Duffy (Head of Product at CalypsoAI) about how an AI model can fail when deployed into a real-world environment and how VESPR Validate is a user-friendly way to identify the impact of fog, motion, etc on your AI model’s effectiveness. 

Request a demo to learn how CalypsoAI’s VESPR Validate will accelerate your AI missions.

Quote:AI model validation is the exercise of giving the people who understand the operational needs, or the mission needs into which an AI model is going to be deployed, giving them the data to understand the model vulnerabilities in their context; to understand, where’s the model going to improve outcomes, or where the model weaknesses are potentially going to introduce new risks or a different type of risk. What we’d like is for those people to be able to interact with the data science development teams or the data engineering teams and with the leaders within those organizations, and ensure that the right model is better for the right task, and that where that model isn’t going to stand up to the various rigors of the production environment, that workarounds can be put in place, or that the business is ready for that and has understood and accepted that ahead of time.– Barry Duffy