Making sure your AI does the right thing: Ethically, legally and financially

NOTE: THIS IS AN EARLY TIME SLOT SESSION - repeated on Thu Oct 6 at 3:30-4:30 pm ET.

It may be sad news for the data science cowgirls and cowboys: But the Wild West days of AI development are clearly over. And that may actually be a good thing as companies and individuals are realizing an old truth: With great power comes great responsibility.

The motivation to use AI in more responsible, more controlled ways is based on multiple sources: Ethical and societal responsibility as a good -corporate- citizen is a base. But in many countries there are now legislations -active and upcoming ones- that will legally mandate a level of scrutiny and responsibility. And many companies also realize that using ungoverned AI is just a bad business decision: Evaluating AI development processes and monitoring for metrics like quality, fairness and bias allows to manage risk and ensures long term success.

But doing the right thing is neither easy nor cheap: Establishing AI governance processes and tools can bloat bureaucracy and inhibit innovation. This is where tooling can help to balance oversight and control with productivity and innovation.

In this session we will discuss the drivers for AI Governance, the challenges that companies face in “doing the right thing” and we will show some of the tools and processes we have come up within IBM to meet these challenges.

Presenter: Thomas Hampp-Bahnmüller

Thomas Hampp is the Senior Technical Staff Member (STSM) for AI Governance in IBMs Trusted AI area. He led the productization of the Model Inventory and AI Factsheet functionality in this area working closely with IBM Research on this. He works in IBM's software development organization and is responsible for the technical architecture and integration of AI Governance/Factsheets with the relevant products in this area like Watson Knowledge Catalog, Watson Studio and Watson Machine Learning and Watson OpenScale.

In previous roles, he was responsible for Information Lifecycle Governance (ILG) development including eDiscovery and archival. Thomas has published several papers on governance technology topics like “Infrastructure requirements for an efficient implementation of information governance in enterprises”, “Improving the Efficiency of Legal eDiscovery Services using Text Mining Techniques” or “Automated Concept Extraction to aid Legal eDiscovery Review”.

Thomas is a well-recognized expert in the domain who holds several patents in the area, he has been actively involved in standardization of key software interfaces at OASIS and has been teaching guest lectures at universities and been an active speaker in industry conferences. Thomas initially joined IBM in the Research division where he was working on machine learning for content classification, search, and analytics. Thomas holds an MA for Computer Science/Computational Linguistics from the University of Tübingen.

***

It is recommended that you register at this Webex link ahead of time to receive a calendar invite and reminder. https://ibm.webex.com/ibm/j.php?RGID=r5ef428d09afe686c0f8489e016c9d2b0