Adopting, Advancing, and Adapting to AI
Artificial intelligence (AI) has become a nearly ubiquitous aspect of modern life. We interact with AI on a daily basis through autocorrect and predictive text, facial recognition to unlock our phones, the algorithms that determine what shows up on our social media feeds, and traffic predictions in navigation apps. It’s already hard to imagine life without these modern conveniences. More broadly, AI is changing the job market in ways we cannot predict, which presents opportunities and challenges for institutions of higher education.
Science fiction writers and Hollywood have provided the blueprint for the future of humanity and its workforce. On one end of the spectrum, it imagines a utopian lifestyle where AI does all the work and humans have space for creativity and leisure as in the Disney/Pixar movie Wall-E. On the other, perhaps a self-aware AI that dominates the world at severe cost to humans as in James Cameron’s Terminator movies. The reality may be neither of these extremes, and universities like Penn State are finding ways to prepare students for as-yet unknowable future careers, and our faculty are transforming how they approach research.
In the Eberly College of Science, researchers are leveraging AI to manage massive datasets, identify patterns in complex images, evaluate tools to improve human health, test the properties of newly developed materials, and a myriad of other applications. Instructors are finding innovative ways to incorporate AI into their classrooms to support student learning and to equip them with the skills and knowledge needed to thrive in an AI-driven workforce. But what is artificial intelligence, and are people really talking about the same thing when they say AI?
What is AI?
Artificial intelligence loosely refers to machines with human-like “intelligence.” This can include qualities like learning and problem-solving, but what constitutes intelligence—and the markers of that intelligence—is still up for debate.
The field of AI research officially originated at a workshop of researchers at Dartmouth College held in 1956, though ideas of artificial humans can be found throughout history and myths, arguably dating as far back as the ancient Greeks. The meaning has changed over time, sometimes referring to specific algorithms or popular generative AI models and sometimes to an anxiously anticipated future self-aware machine that will bring with it a horde of ethical, security, and other challenges.
Today, AI is largely an umbrella term for the multitude of increasingly prevalent and rapidly evolving technologies that are shaping how we do science, how we teach, and just about everything else in our daily lives. These technologies, from machine learning to neural networks, all rely on modeling uncertainty and use theory and methods from mathematics and statistics. Although the terminology and technology are changing by the day, it’s clear that AI, in some form or another, is here to stay.
AI and Society
The rapidly advancing field of AI has brought with it a lot of excitement, but also apprehension. Much like the advancement of the internet, which had rapid early development and has changed dramatically over the years, AI offers potentially huge benefits, but we need to navigate this thoughtfully. AI could help us manage the deluge of data that other advancing technology is allowing us to collect and store at an unprecedented rate in fields ranging from health to astronomy. It could help us innovate and mitigate ongoing crises, for example climate change through improved climate models and optimized energy grids, and has already made huge strides in accessibility, for example by generating alternative text and subtitles on images and videos.
However, more research is required to understand and inform concerns ranging from the ethics of sourcing training data to the impact on our own brains. AI is here to stay, and there is work to be done to make AI more efficient and sustainable, to clearly identify when it is—or isn’t—appropriate to use, how that can be done ethically, and to prepare students for an AI-driven workforce.
“Artificial intelligence on the whole has already been incredibly transformative, and we continue to explore how to harness this technology to enhance our research, teaching and operations in the Eberly College of Science,” said Tracy Langkilde, dean of the Eberly College of Science. “Our researchers have been using machine learning and other forms of artificial intelligence for decades to make advances in everything from understanding our bodies and environment to materials and even the cosmos. Now, it is allowing us to do things we never thought possible.”
AI and machine learning—which falls under the AI umbrella—have supported research across all departments in the college. Additionally, some of our researchers are working to develop, improve, and better understand artificial intelligence methods, theory, and algorithms. Many of our researchers are involved in and have leadership positions within Penn State's Institute for Computational and Data Sciences AI Hub, which connects interested researchers from across the Penn State system, as well as its many centers, including the Center for Socially Responsible Artificial Intelligence.
Our instructors are also thoughtfully experimenting with how AI can successfully be used as a critical tool to transform teaching and learning while maintaining the integrity of the educational experience. In addition to launching new AI options for undergraduates majoring in mathematics or statistics, the college is working to improve AI literacy in our community and especially among our students, with the goal of teaching students how to use AI responsibly, creatively, and effectively.
“In the age of AI, Penn State has the opportunity to define what it means to be a world-class land-grant institution, including advancing human-centered innovation, ethical leadership, and lifelong learning,” said Fotis Sotiropoulos, executive vice president and provost at Penn State. “The strategic focus areas we’ve identified aim to build career-critical AI skills and AI literacy at every level, ensure that every student thrives in the age of AI, position Penn State as a national leader in AI-driven research and innovation, and build a modern digital ecosystem.”
In this feature story, we will dig into the various types of AI and explore a wide variety of ways members of the college are developing and using AI in innovative ways to support their research and teaching.
The various forms of AI each have their own pros and cons, which are worth considering when identifying the most appropriate tool to use.
Machine Learning and Artificial Intelligence in Research
A metic named after a Penn State professor supports many AI tools.
The Kernel method provides an alternative to deep learning.
What is deep learning, and how can mathematics improve deep-learning models?
Graduate student Maxwell Konnaris aims to improve AI and other tools for genomics.
AI helps researchers determine structure of “dancing” molecules.
New grant supports creation of "digital twins” to study disease progression and treatment options.
How AI can help researchers processes ever-increasing amounts of data.
Identifying new materials could help with search for elusive dark matter.
Building models for star variability using physics-informed machine learning trained on the sun.
How Can Deep Learning Improve Physics and Engineering?
In this Q&A, former Penn State postdoctoral scholar Robin Tuluie, now CEO of PhysicsX, describes how deep learning can be responsibly used to transform physics and engineering.
