From explaining model behavior and mitigating bias to preserving data privacy and tracking model lineage, Azure Machine Learning’s state-of-the-art technologies help users develop and maintain ML responsibly. Our offerings will continue to expand and evolve as the field advances.
It is worth noting that tools for data scientists and developers are not be-all-end-all solutions. For long-term success, organizations must use technical tools within a more holistic approach to AI. That approach will be unique for each organization, but there are a few common threads. For instance, in addition to using tools like the ones in this paper, technical teams should establish guidelines, checklists, and processes that foster ethical practices, rigorous testing, and continual auditing.
Organizations should have robust systems for governing AI and maintaining accountability. IT teams should be trained in ML-related skills, and all employees should be trained to understand ML solutions and use them appropriately. IT teams should collaborate with their business colleagues to determine the best data and training techniques to use for achieving the desired
outcome for each model. To learn more, download this paper and visit the Machine Learning practitioners page at http://aka.ms/data-scientists.