How to Make the Most of AI
The advancing field offers valuable tools for educators and growing opportunities for research.
By Aditya Johri
For millennia, humans have developed tools and technologies that augment our abilities and help us perform more effectively. As engineering educators and researchers, we have come to rely on information technology (IT), which shapes our own practices and those of our students. The swing toward online education and remote work driven by COVID-19 has only hastened the transition to technology-driven work practices. Our most advanced IT tool currently is artificial intelligence, which has slowly started to augment, if not replace, human practices. By driving the products we use, AI algorithms change how we work.
Most AI experts agree that artificial general intelligence (AGI), the kind of AI that can work just like the human brain, is a long way off. However, artificial narrow intelligence (ANI) is making rapid progress. Although ANI has limitations and biases, its applications—driverless cars, drone deliveries, and conversations with Alexa—were unfathomable except as science fiction just a few years back. By offloading our thinking and exchange of ideas to an external representational system—a language—ANI enables us to handle complexity we might not otherwise have tackled.
In the near-term, the engineering education community can use AI to make a meaningful impact in three areas:
The first is data. The quality and quantity of data that we generate across teaching, research, and administration have changed. They need to be managed, analyzed, and reported in better ways to improve engineering education. As researchers we generate far more data than we are using effectively, and sharing of data is essential to insights that are more nuanced, not just generalizable. It is important to be able to combine learning data, coming from a learning management system, with financial data and advising support. From the world of AI, machine learning (ML) and, more broadly, data mining (DM) approaches can help us do a better job.
Second, content is omnipresent, providing both opportunities and challenges for how to deal with its vast amount and diverse forms. Curation is a difficult issue but also one of the biggest benefits, if utilized, of the Internet, and AI techniques can assist. Through a combination of data mining approaches and expert sourcing, crowdsourcing, and student sourcing, we can generate more effective ways to organize content. On online communities, such as Stack Exchange, GitHub, and Reddit, experts share knowledge, and over time these repositories themselves become a source of information and can assist with curriculum development. They also provide worked-out examples that can be incorporated across teaching contexts.
Third, workforce automation is changing how engineers work with each other and with machines. As machines increase their ability to do more processing, store more data, and run more complex algorithms, the augmentation of the work of engineers will change correspondingly. We might not be able to predict where all this will go, but we can start trying to understand the AI-augmented world of work practices better, and this might require new ways of researching. This knowledge will be crucial in preparing the future workforce.
With the opportunities of AI, though, come risks. A cross-cutting ethical theme is the problem of algorithmic bias. Biases of different sorts, including of gender, socioeconomic status, and race, are increasingly being cemented into systems. The sources of many of these biases are the data that undergird the software design. And unless better data are collected and then used to design systems, biases will perpetuate. Preventing biases and mitigating the negative outcomes associated with them are important concerns for engineering education, especially as the field is still struggling with institutionalized discrimination. Inclusion of AI systems within the engineering education workflow can further exacerbate these problems, rather than helping us find solutions.
We also need to be vigilant about so-called black-box technology that is not open to scrutiny.
We don’t need to jump overzealously on the bandwagon of AI when it comes to engineering education. At a time when just being able to provide instruction to students can itself be tricky, adding a new ethics topic to the curriculum might seem daunting. Yet we cannot afford to ignore AI. Its impact is such that how we respond will determine the progress we make both in the classroom and in reaching social justice goals of access and equity.
Aditya Johri is a professor of information sciences and technology at George Mason University. This column is adapted from “Artificial intelligence and engineering education” in the July 2020 Journal of Engineering Education.
Image Courtesy of Creative Services/George Mason University