Abstract: 

It is unquestionable that, AI, machine learning, Big Data and data science have recently gone through a new inflection point. Whereas in the past, data science was rooted in two communities--statisticians (or derived fields) and computer scientists--its future is already showing that it is branching away from these origins.

What was later called “data science,” clearly originated from mathematics. In those early days, most work in the field was focused on mathematical properties of methods it relied upon and their optimal characteristics. Practitioners were then experts in understanding fundamental assumptions of models and their interpretability. With the explosion of our compute capacities and the size of data we are able to handle, the field needed the impetus of computer scientists to take it to another level. The core concern was no longer optimality or properties but scalability; “data trumps algorithms” and (prosaically speaking) area under the curve (AUC) took over limit theorems.

With the outsized impact that Data Science now has on human society, it is time to step back from our AUCs and take a hard look at the Deontology of Data Science. While there is a generally understood and at times very mature Deontology in professional communities that interact with “Humans”, it is unfortunately not yet present at all in our field. How can we attempt to understand the Deontology of our field amidst its potential and contemporary social norms, business motivations and potential impact on all of us?

Bio: 

Igor Perisic is the Chief Data Officer and Vice President of Engineering at LinkedIn. During his tenure, he has led team that has developed LinkedIn’s Search Engine, its Real-Time Graph Engine, Relevance infrastructure and worked on personalizing LinkedIn’s site for its members.

Igor and his team contributed back to open source technologies with projects such as Kafka, Voldemort or extensions such as ml-ease and many others.