Artificial intelligence (AI) started off with great promise in the early 1980s, spurred by the success of the expert system paradigm in certain applications. This prompted a flurry of research activities in chemical engineering in the mid-1980s. However, as the ensuing three decades showed, AI didn’t quite live up to its promise in chemical engineering. So, what went wrong with AI?
In this talk, I will review the different phases of AI in chemical engineering over the last 35 years, providing some background and explanation to this question. I will also argue that this time it is different – I believe the time for AI in chemical engineering, and in other domains, has arrived, finally. There are many applications that are ready to yield quick successes in this new data science phase of AI. I will highlight recent work in materials design and in process operations as examples of exciting progress.
However, the really interesting and intellectually challenging problems lie in developing such conceptual frameworks as hybrid models, mechanism-based causal explanations, domain-specific knowledge discovery engines, and analytical theories of emergence. These breakthroughs would require going beyond purely data-centric machine learning, despite all its current excitement, and leveraging other knowledge representation and reasoning methods from the earlier phases of AI. They would require a proper integration of symbolic reasoning with data-driven processing. I will discuss these challenges and opportunities going forward.