Marketplace model offerings should endeavor for transparency and explainability, in order to increase trust in AI by end users and other stakeholders. Allowing a clear understanding of how the model functions is key to ensuring the end user can identify the right opportunities for use, trust in the outcomes or predictions, and mitigate the effects of bias as it is detected.
Marketplace model offerings should perform to declared standards and not cause unintentional harm. Model providers are required to submit verifiable metrics on performance, as well as to furnish relevant technical and training data descriptions so that end users can make informed choices to enable the missions or business needs they serve.
AI should aspire to prevent harm; respect for privacy and protection of data on individuals are fundamental to fulfilling this requirement. This includes not only adherence to applicable laws, but ethical employment of data derived through technological processes.
The Modzy Marketplace should enable diversity and fairness through inclusive, equitable, and appropriate model offerings that respect the rights of all. Modzy will not tolerate any model offerings that are not lawful and beneficial to society.
Marketplace model offerings should be designed and perform in a way that provides governance and auditability. This necessitates appropriate controls and performance monitoring mechanisms in order to detect shortfalls in performance or the emergence of issues with fairness, privacy, or security.
Get a video demo of Modzy, the ModelOps and MLOps software platform that businesses use to deploy, integrate, run, and monitor AI—anywhere.