Biases in AI systems

Where: Zoom Join from PC, Mac, iOS or Android: https://vuw.zoom.us/j/96774938033

When: 10–11 am Friday 29 April 2022

Bias in AI systems is a growing concern across a variety of industries.  Despite considerable research effort many bias related challenges remain. Creating   high   quality datasets, development of   efficient fairness-focused auditing tools, and creation of accessible educational resources for non-AI experts (such as software engineers and business executives) to understand bias are just  some  of the  many  challenges that  need  to  be addressed. The broad goal of this talk is to serve as an accessible guide in educating software engineers about types  of bias,  to provide them with  a checklist  of  biases that they  need to potentially test for, and list some guidelines that could aid them in assembling various stages of a robust AI pipeline. Specifically, a taxonomy of various types of biases that can creep into the AI pipeline, from the data creation and problem formulation stage to the data analysis and evaluation stage will be discussed.

Biography:

Dr Ramya Srinivasan is an AI researcher in Fujitsu Research of America, where she is a member of the AI Ethics Research Center. Ramya’s background is in the areas of computer vision, machine learning, explainable AI and Ai ethics, with her research spanning diverse application areas such as arts, healthcare, and finance.

Contact:

Phone:

Email: