As data and artificial intelligence come to play an ever larger and more important role in business and society, many non-tech companies and organizations are asking how they might be able to put AI and machine learning technologies to use in their line of work. Professor Eric Yu and his team of researchers, newly funded by the Natural Sciences and Engineering Research Council (NSERC), are working to help provide some answers.
Their research focuses on allowing organizations, ranging from for-profit companies to government departments, to understand how AI and ML technologies could fit into their processes and improve their operations. The latter could include better understanding of clients and consumers leading to improved interactions, developing more effective equipment maintenance systems to minimize failures, and analyzing data to make predictions about the future.
“Leading edge companies are doing all of this,” says Yu, pointing out that the financial and retail sectors are well advanced in their use of AI and ML. “Many others are rushing in because these technologies are all the rage, and if they’re not using them, they feel they’re out of the game.”
Professor Eric Yu’s earlier modelling work has become an international standard.
But while the technologies are out there, as are the people capable of building them, many organizations don’t know how to think about AI or how they could actually use it. “They don’t yet have ways of conceiving how artificial intelligence and machine learning can fit into the organizational environment,” explains Yu. “That’s what this project is about.”
Officially titled “Requirements Modeling for Enterprise Applications of Machine Learning and Artificial Intelligence,” the five-year-long project aims to develop modeling techniques that would allow organizations to be able to take advantage of AI and ML in much the same way they now reap benefits from conventional information systems.
“These kinds of models are well established for conventional information systems,” says Yu, citing financial and medical records as examples. While a conventional information system is used to store, search and access these records, AI and ML systems can now use such records data to diagnose diseases, come up with treatments and reduce costs among other things.
In the transport sector, ride hailing companies are using sophisticated AI technology to control everything from pricing to the whereabouts of their cars but many large public transit organizations, who could also benefit from AI to do things like make better route predictions and save money on gas lag far behind. They are just one example of companies that could be helped by this type of research.
“We’re going to do it iteratively – start with conceptual developments, come up with ideas how to model, test with case studies, seek out organizations to test our methods, improve our methods, and get wider feedback through publications,” says Yu. “At the same time, we will try to promote these methods so they eventually get adopted.” (Some of Yu’s earlier modelling work – the so-called i* framework, which is the basis for the Goal-oriented Requirements Language (GRL) – has become an international standard.
In some cases, the changes spurred by AI will be visible and dramatic, says Yu, while, in others, they will be imperceptible or may just appear as quality improvements.
With more activities going online and becoming digitized during the pandemic, there’s more data being generated providing more opportunities for organizations able to put data to use. AI systems “feed on data,” says Yu. “Without data there’s nothing these AI systems can do.”
