(449e) Teaching Artificial Intelligence to Chemical Engineers: Experiences from a 35-Year-Old Course

Given the recent success of artificial intelligence (AI) in a wide variety of applications, it has become quite clear that chemical engineering students need to become familiar with applied AI techniques [1]. There is great interest in the academic community in developing a course to address this important need. I have been teaching such a course to chemical engineering undergraduate and graduate students since 1986 [2], first at Columbia for three years, then at Purdue for about twenty years, and back again at Columbia now. It was mostly on symbolic AI in the early years, with an emphasis on knowledge-based expert system concepts and techniques, for various applications in chemical engineering such as design, diagnosis, control, optimization, and process safety. From the 1990s onwards, as machine learning techniques started showing promise, my course evolved to include topics such as clustering, neural networks, statistical classifiers, graph-based models, and genetic algorithms. The current version at Columbia has the first one-third of the course on symbolic AI and the rest on machine learning-based numeric AI concepts and techniques. It is mainly a project-based course where the students develop systems using AI techniques for chemical engineering applications. In the early years, the students programmed in AI languages such as LISP and OPS5, but now using the easier tools such as Python and TensorFlow. In this presentation, I will discuss my experiences in teaching this course, including the syllabus and the projects we have done over the decades.

Reference:

1. Venkatasubramanian, “The Promise of Artificial Intelligence in Chemical Engineering: Is it here, finally?”, Perspective article, AIChE Journal, 65(2), pp. 466-478, 2019.

• Venkatasubramanian, V. “Artificial Intelligence in Process Engineering: Experiences from a Graduate Course,” Chemical Engineering Education, Fall Graduate Education Issue, November 1986, pp. 188 – 192.