I just attended a workshop for college presidents about the impact of Artificial Intelligence, or AI on higher education and the responsibilities and opportunities it presents our institutions and society. The benefit of this week’s workshop was being able to interact with leaders in the industry and colleagues trying to position these same issues on their respective campuses.

AI Background

AI was established as an academic discipline in the 1950s. It built on Alan Turing’s pioneering work, which was the foundation of machine learning. AI is fundamentally a set of tools that allow us to navigate and manipulate large sets of data and to generate new material based upon powerful machine learning with the ability to sift and sort enormous sets of data for source material.

Most of us benefit from AI applications that are embedded in more and more of the technology we use. Search engines and the curatorial features of streaming services and e-merchandizing use various forms of AI.

If you have swiped up on a photo on your smart phone to identify an image, that image recognition used AI. Auto-complete functions in texting, email, and word processing are driven by “Language Learning” technology, as are editing programs like Grammarly. Self-driving and semi-autonomous vehicles utilize AI technology.

AI is dependent on computer hardware with powerful math processing capabilities. Graphic Processing Units (GPUs) were originally created to support rich visual content for gaming, but eventually, enterprising graduate students realized that the speed and capacity of these units could be used to solve complex computing problems. Many of the most significant developments in this technology have been led by NVIDIA. They were the hosts for this week’s meeting,

Some Current AI Technologies

Image recognition ranges from the swipe up feature on your phone to sophisticated facial recognition technologies. There have been many ethical challenges from biased data sets and the implicit bias of programmers, but this is an area where computers are now outperforming humans.

A generative AI complement to image recognition is text to image technology, which will synthesize an image based upon a text description. The initial image can then be edited using text prompts as well.

One of the machine learning functions that has been most in the news lately is Large Language Modeling (LLM). The 2022 release of ChatGPT (a chat bot that uses Generative Pre-trained Transformer technology, which allows the program to do much of its own training) introduced this functional technology to tens of millions of people almost overnight and pushed AI into the spotlight.

ChatGPT gave us open access to technology that could pull data together in an instant and synthesize textual summaries. It could also set the style and audience level with surprising sophistication.

On college campuses, as with all topics, there were strong and divergent views. Some faculty immediately worried that students would use this technology to “plagiarize” original work, others saw it as a tool to help students improve their writing with its capacity to rewrite large passages of text in seconds. Still others were taken by its ability as a powerful search engine with the ability to contextualize and curate the materials it gathers.

It didn’t take long for many of us to realize that it makes mistakes. Some of this is the result of the fallibility of its data sources, but it also fabricates and “hallucinates.” Much of the language learning design is focused on its ability to write clearly and compellingly, and this seems to generate material that is either untrue or unreal. These aberrations are prevalent enough that ChatGPT and Gemini have disclaimers that they make mistakes. Over time, these problems appear to be reducing, and they are likely to be worked out in the coming years.

Another function of AI uses a Digital Twin, which is a digital “copy” of a real thing or a proposed model. This technology makes advanced simulations and testing possible. Thousands of iterations of a science experiment can be simulated in a short timeframe speeding up research, reducing costs, and expanding data collection. Likewise, designs and models can be tested in accelerated time to improve functionality and safety and reduce the costs of development.

Many of those functions are dependent upon High Performance Computing. These often use supercomputers or computer clusters to process enormous sets of data quickly.

The Role of Higher Education

We describe a Susquehanna education as one that makes our students “Future Ready.”

Part of this is our recognition that the soft skills rooted in the liberal arts prepare students to be adaptable to an ever more rapidly changing world. We also offer a curriculum that provides a distinctive integration of those evergreen foundations with career-oriented skills. Those skills now need to include being articulate about AI, having the facility to use it as a problem-solving tool, and understanding the moral and ethical framework in which it is being used.

As one of the participants in our workshop said, “While we’re figuring out what to do with AI, our students are graduating into jobs where they are using it.”

Another added, “AI won’t take your job, someone who knows AI will take your job.”

AI displaces the need for certain skills. This will eliminate some jobs, but it will alter the focus and scope of many more. Some skills will be more needed and some less. AI has the potential to democratize many work-place skills

I recently formed a task force of faculty, staff, and students at Susquehanna to examine these issues and to set policies and practices for our campus, so we can make the most of these emerging technologies. This will position us to be proactive in preparing our students for the emerging world of work with AI, and it will help our campus take better advantage of the operational and research opportunities these technologies have to offer.

Business as usual is no longer an option, because the playing field is no longer the same.

Instead of merely using AI to improve how we play the game, we need to think about how we can change the game?

One of our sessions was about how the University of Florida has implemented AI across the curriculum. Not everyone will be an expert, and it is not a graduation requirement, but students in every program have the opportunity to learn how AI works, and what its current and emerging applications are in their respective disciplines. Students also have the option of completing a 3-course certificate program.

We are on the cusp of the Fourth Industrial Revolution, which will be marked by dramatic increases in automation, a transformation in human-machine interactions, and significant shifts in where economic power sits.

We have the opportunity to shape the narrative of where this revolution will take us, how we can use it to address some of our most intractable problems, and how we can use these technologies to support a fairer global society over time.

The time for those first steps is now.