Shaping the future of AI: A Q&A with Rada Mihalcea

For over 35 years, the Artificial Intelligence (AI) Lab at the University of Michigan has been a leading hub for AI research, fostering innovation and informing the direction of the field. First established in 1988, the AI Lab has consistently pushed the boundaries of what’s possible in AI, harnessing the ever-growing power of intelligent systems to advance the field and better society.

From its early days under the leadership of founding director Ramesh Jain, who left U-M in 1993, the AI Lab has grown in both scope and impact—venturing into cognitive architectures, natural language processing, healthcare computing and more. 

As the AI landscape has expanded, so has the lab, addressing some of the most complex and pressing challenges in the field. Today, with over 25 faculty and research staff and nearly 100 PhD students, it is one of the nation’s largest academic AI labs, working collaboratively with industry and with experts from other disciplines to drive groundbreaking advances.

We recently sat down with Rada Mihalcea, the Janice M. Jenkins Collegiate Professor of Computer Science and Engineering and the director of the AI Lab, to learn more about its trajectory and ambitious vision for the future.

Can you give us a big picture overview of what the AI Lab is trying to accomplish?

Our main goal is to advance the fundamental technologies of AI, covering core areas like machine learning—including deep learning and reinforcement learning, computer vision, natural language processing, and speech processing, along with related areas such as multi-agent systems and cognitive architectures. Our work involves both theoretical exploration and practical application, tackling problems like decision-making and systems of interacting agents, and extends to areas like healthcare and robotics.

AI technologies that are seeing more publicity and widespread use today, like large language and vision models, are just a slice of the field.

While we certainly engage in applied research, the emphasis is on foundational work to drive advances in methodologies and technologies. This ensures that we’re continuing to innovate in these areas, rather than relying on current capabilities, which could become obsolete in a few years. We’re also focused on identifying and addressing the limitations of present models, which will ultimately allow us to unlock even more areas where AI can make an impact.

What are some of the most exciting areas of research happening at the AI Lab?

We have specialists in many areas of AI, and really groundbreaking work is being done in all corners. One exciting avenue is improving AI models to function effectively in low-resource environments, broadening their accessibility while also lowering their environmental impact. We have faculty working to improve the accuracy, reliability, and trustworthiness of AI models, as well as leveraging natural language processing to help machines better understand and respond to human language.

There’s also significant work being done to understand these models’ internal mechanics to address challenges like hallucinations or biases. For example, researchers are exploring how biases develop in neural networks and investigating methods to mitigate them. Others are examining the mathematical structures behind machine learning algorithms to predict how models generalize from training data to new, unseen data.

A major focus of my own research is in the realm of AI for social good. I spearheaded the NLP for Positive Impact (NLP4PI) initiative, which aims to tackle pressing social problems by leveraging NLP technologies. Our goal is to empower researchers in natural language processing to expand the societal impact of their work by providing a rich set of resources and a network for collaboration.

Our lab’s strength also lies in collaborating across domains, including healthcare, social sciences, finance, robotics, and more. We have researchers working collaboratively to develop AI-powered assistive technologies for individuals with disabilities, create models that can help monitor mental health symptoms, enhance predictive models for medical diagnosis, and much more. These collaborations not only expand the applicability of AI but also ensure that the solutions we develop are deeply informed by domain-specific knowledge, leading to truly impactful advances.

With AI technology applications becoming more widespread, why does fundamental AI research still matter?

Fundamental AI research is critical for ensuring continuous progress and not getting stuck with technology that only makes incremental technical improvements from one year to the next. AI technologies that are seeing more publicity and widespread use today, like large language and vision models, are just a slice of the field. By investing in foundational research, we can continue expanding, understanding, and exploring more nuanced facets of AI. 

For instance, beyond simply developing new models, it’s vital to address their limitations—such as reducing biases, preventing hallucinations, and enhancing fairness and interpretability. These improvements are essential for creating AI systems that can be reliably and ethically used in society.

By investing in the fundamentals of AI, we’re not just refining existing technologies—we’re paving the way for applications that have the power to transform industries, improve lives, and address complex global challenges.

In addition, fundamental research expands AI’s applicability to a wider range of domains and populations. It enables us to tailor AI technologies to meet the specific needs of diverse users from different backgrounds and fields. 

By investing in the fundamentals of AI, we’re not just refining existing technologies—we’re paving the way for applications that have the power to transform industries, improve lives, and address complex global challenges.

How is the AI Lab fostering interdisciplinary collaborations, and why is this important?

Interdisciplinary collaboration plays a vital role in creating AI solutions that are both innovative and practical. In the AI Lab, we partner with healthcare professionals, social scientists, and many other domain experts across the University of Michigan and at other institutions. For instance, our collaborative initiative with Michigan Medicine, e-Health and Artificial Intelligence (e-HAIL), exemplifies how balanced contributions from AI and healthcare experts lead to successful outcomes. Recent projects to come out of e-HAIL include the development of an AI-powered reproductive health chatbot, and an effort to leverage AI tools to improve the diagnosis of dizziness.

We also have researchers in the AI Lab working with colleagues from the School of Information, and those in fields like psychology, law, economics and business, engineering disciplines, and more. These collaborations are crucial for integrating varied expertise and insights into our AI research. Engaging domain experts from a project’s inception allows us to develop robust solutions that maximize impact, ensuring that AI applications are not only technologically sound but also beneficial in real-world settings.

What unique strengths does the AI Lab have that help it stay at the forefront of AI research?

One of our primary strengths is the breadth of core AI areas we cover—encompassing nearly all facets of AI—which is relatively rare among universities. Our lab is home to around 25 talented faculty members and numerous PhD students, facilitating a strong interdisciplinary and collaborative environment. Our consistent presence in top AI conferences and global rankings as a leading AI program testify to this excellence.

Beyond academic accolades, another factor that sets us apart is our prioritization of social impact, addressing significant issues such as healthcare, climate change, poverty, and mental health through dedicated initiatives. This multifaceted approach allows us to not only advance AI research but also apply it in meaningful ways that can significantly affect people’s lives.

We’re also proud of the broad representation within our lab, which truly enhances the depth and reach of our research. These elements ensure we can tackle a wide range of AI challenges and innovate effectively.

Looking to the future, what are you most excited about at the AI Lab?

I’m excited about the potential for AI to benefit society through human-centered solutions. I think we will see an increasing focus on developing AI that genuinely enhances human life and addresses real-life challenges, aiding in decision-making while being mindful of ethical concerns and privacy issues. Looking ahead, much work remains in advancing learning and reasoning, language processing, computer vision, and cross-disciplinary applications of AI. My vision for the future is one where AI naturally integrates into daily life—and does so effectively for those of different backgrounds and abilities—assisting us in tackling complex challenges and making a positive difference for individuals and communities alike. AI has the potential to transform society for the better, and that is what drives our research.