Abstract:
Today, there are more than fifty domains in which AI programs perform at least as well as humans, and there are at least a thousand more that are helping humans in making better decisions. All of them require data, without which the underlying software would be simply a bunch of algorithms and not accurate models. Undoubtedly, just like oil fueled the industrial revolution, data will fuel AI. Hence, data has been often called the “new oil”. However, this analogy obscures the multifaceted nature of data. For example, a given amount of oil can be only consumed once whereas the same data remains undiminished after being used several times and for several different use cases. This talk discusses the following complex facets of data: • More noise-free data often beats better algorithms and results in more accurate AI models. • Role of bias in data and a few domains where such bias is particularly harmful. • Ways of mitigating bias in data (although this like removing bias from human society). • Tricky aspects related to ownership, consent, purpose, privacy, security, and auditability. We conclude by discussing the intermixing of these issues with training of inexplainable AI systems, which eventually need to be fair and ethical.

Abstract:
With tools, technology and computational power available to us today, there is seemingly no limit to the amount of data that can be leveraged to get to improved decision making and business value. However, we have learned the hard way that there are limits when it comes to business adoption of data science and artificial intelligence solutions.  From lessons learned of getting caught up in the hype of artificial intelligence to major success stories, this talk will explore real-world examples of Southwest Airlines’ data science journey over the last several years.   

Abstract:
Artificial Intelligence (AI) and Machine Learning (ML) applications have pervaded all aspects of our lives, ranging from systems that recommend products to those that diagnose diseases that may defy the best of doctors. However, questions about the trustworthiness and reliability of such systems abound, as many of the AI/ML applications do not explain the rationale for a prediction or a recommendation. The widespread acceptance of such applications depends on moving from black box models to explainable AI (XAI) systems that facilitate traceability, interpretability, and accountability of AI/ML applications. In this discussion, our panel of experts will discuss the consequences of a black box approach and offer their insights into how organizations can build trustworthy XAI systems.

Abstract:
As organizations proliferate Artificial Intelligence (AI) and Machine Learning (ML) applications to draw actionable insight and unlock business value, there is growing concern regarding issues such as fairness, bias, privacy, and ethics related to the use of such applications. In this panel discussion, our experts will discuss the ethical and social dilemmas that these applications present as well as the consequences of deliberate or unintentional data and algorithmic biases that may be introduced while building AI/ML systems.