Approximately 28 years ago the department chair and a junior in his department entered into a debate. The style of debate wasn’t uncommon for a liberal arts campus and the professor seemed to cherish the opportunity to discuss a topic dear to his heart. The professor had taught multiple classes to me, the student, since freshman year at Central College. That common area of the Math/CompSci department had hosted many such conversations and faculty were well aware of my preference for applied computer science. My disdain for the abstract and theory were well-known. The debate lasted a couple of hours and ended in stalemate.
Fast forward to this week when I spent a few days at an artificial intelligence conference hosted by O’Reilly. The conference was hosted by a book publisher focused on applied programming – from languages to systems, from the esoteric to the mainstream. A few hundred people came together to learn the concepts and applications in artificial intelligence to produce AI systems capable of machine learning.
It was a keynote by Peter Norvig , a director of research at Google, that reminded me of the three decade old debate. In the keynote, Peter spoke about the changing paradigm of programming from where we teach machines to “DO” to where we teach machines to “LEARN”. The learning achieved through probabilities assigned to outcomes. The binary trees of outcomes with a statistical probability. The literal mashup of Comp/Sci theory and statistics. I realized then why I’d struggled with the workshops and some of the sessions. I was fine understanding the SQL queries and python code but was lost when it came to understanding the probabilities assigned to query results and analyses!
So, Dr. Meyer, you win. Our debate was about whether a Computer Science degree was earned without classes in statistics. You maintained that it wasn’t as Statistics was a core member of the curricula. I disagreed and seemed to remain right for 26 years in the profession. If I am to proceed any further into the reaches of AI and machine learning, however, I must first begin with a mea culpa and then put my liberal arts education to work and reopen the book on statistics before delving further into AI and ML.