Ben Joseph, Chief Data Officer and AIG at USPS OIG, speaks with Michael Hughes, Chief Business Officer of Duality Technologies, in a video interview about complying with regulations, following data lineage, ensuring transparency with AI and ML, and the importance of data collaboration.
Speaking about complying with regulations while handling data and leveraging other tools, Joseph mentions following some pre-existing standards. In addition, he refers to following the “Yellow Book: Government Auditing Standards” while handling data for customers like auditors.
Delving further, Joseph states that the data lineage must be followed throughout, since the beginning with the data source. Terming it as data reliability, he says that it is critical for auditors as they need reliable data while writing audit reports.
Moving forward, Joseph states that the principles to assess data reliability are taken and applied to the organizational data governance or data quality principles. When it comes to AI and ML, the organization is coming up with transparency.
To ensure transparency with AI and ML, the organization assesses the models being used, and if they are supervised or unsupervised. Next, if it is unsupervised, the data scientists investigate how the algorithm was created, what powers the model, and if it is reproducible.
While getting into the supervised space, some of the mentioned principles apply, but the data management is well under control, says Joseph. The transparency element is big in the development space, he notes, and to keep things fair, the organization approaches the development side with principles.
Thereafter, Joseph states that including the big tech, nobody has put a clear-cut path and the space is still evolving. For instance, how to approach anomaly detection. Further speaking on the scenario, he states that there is high-level guidance available that needs to be boiled down to specific needs and use tools to make it transparent. It is critical to understand the various tech tools and combine them to define the development space, standardize it, and ensure uniform usage.
When asked about data collaboration, Joseph states that organizations need to create stronger partnerships to understand the value of data collaboration. He maintains that the onus is on the tech people to inform the non-tech people and show them the benefits and value of data collaboration.
Furthermore, Joseph discusses organizing periodic data jams or data hackathons for a year and shares that it has been successful. Through these programs, the organization brought in an entire team and built solutions in a day.
Concluding, Joseph states that the goal is to create a collaborative community for people to understand the value of data and become a part of the data-driven ecosystem.
CDO Magazine appreciates Ben Joseph for sharing his insights with our global community.