AI has people talking as more companies leverage its use and navigate its complexities. See how the folks on South Mountain are integrating AI into their work, learning, and activities.
Craig Gordon ’76
Co-founder of TD Publishing & Research
Adjunct Professor, Journalism
AI Tools: Bing Chatbot, ChatGPT, DALL·E, You, Midjourney
I felt Lehigh was at the forefront of AI in 2016 insomuch as taking a humanistic approach to understanding it and using it with non-tech-centric students. With Jack Lule, Iacocca Professor and chair of journalism & communication and professor of global studies, we presented about AI at the 2019 World Journalism Education Congress in Paris, which was attended by over 1,000 journalism and communication professors worldwide. We stated Lehigh wasn’t just looking at AI from a computer science approach — that’s like having a dessert maker teach people about diabetes. We were bringing a liberal arts lens, but since that time, AI has become a buzzword and media worry.
As a world, we have advanced from an age of enlightenment to an age of entanglement. AI is now multimodal, allowing users to generate text, speech, image, and video. AI can scrape search engines and all other text and images to mimic what is supplied. It still does hallucinate fictional answers, expose implicit biases baked into our systems, and leverage a user’s emotional needs and information sometimes for wrong answers and possible copyright violations. So I can see how people might worry, but we can’t pause its evolution. AI requires guardrails and ethics … that’s again where Lehigh students can help traditional programmers.
I guest lecture in a variety courses at Lehigh to help students understand and apply AI from a non-technical standpoint. We discuss intelligence, history of AI, the types of AI, AI products, pros and cons, and key principles. We focus on how to ask the right questions — creating the prompts that lead to reliable answers. The worry is that AI will answer questions that our reality has yet to consider, questions we don’t even know to ask.
This is why learning AI is crucial as a part of a standard liberal arts education and for non-tech professionals who are in the workforce. People bring wisdom to AI algorithms that currently most demonstrate smarts — it is easy to know the facts and data, but that’s not intelligence … although in some ways it can and is moving fast towards doing that better.
Greg Reihman
Vice Provost, Library and Technology Services
AI Tools: ChatGPT, Amazon Web Services AI, Keenius, Discord, Midjourney, DALL·E
We are beginning to integrate emerging generative AI in three ways: teaching and research, university administrative tasks, and strategic initiatives to position Lehigh as an organization of the future. A cohort of faculty is already partnering with the LTS Center for Innovation in Teaching and Learning to enhance student learning, improve scholarly research, and explore new forms of design using generative AI tools. We are also beginning to train gen AI tools on subsets of information — for example, to support faculty review of grant requirements or to more quickly answer student questions about Lehigh.
We expect these technologies to improve things like the student experience with course registration and students' ability to discover events and clubs that match their interests. There are also countless behind-the-scenes use cases such as helping with classroom planning to take into account professor preference, student needs, proximity to offices or labs, and space requirements, or helping us analyze complex retention data to mine for patterns that lead to student success.
In short, we want to deploy AI tools that keep Lehigh a human-centered, user-focused institution while also empowering our faculty, staff, and students with the enhanced capabilities these emerging technologies provide.
Nik Nikolov
Associate Professor, Art, Architecture, and Design
AI Tool: LookX
While our use of AI is nascent, it has been around a long time and existed in many forms. The future uses for AI are difficult to predict since this tool is now flooding the mass culture and people will begin to innovate and apply it to many tasks. In my classes, students are using AI algorithms to make good predictions, to see how it might impact their creative and decision-making process. There are five prototypes for a dwelling: single-family detached home, single-family semi-detached home, townhome, multi-family residential, and mobile homes. Students use LookX to select certain precedents, like type of dwelling, style, and feeling. They can upload a sketch as well. What is generated provides them the opportunity to critique and edit. They can narrate the production of an image and then speculate the plans needed to bring it to life.
Where we as librarians are right now with AI is similar to what we experienced when Wikipedia became popular or when the internet launched. We have to teach students how to use it and how to cite it. But users have to critically evaluate the information provided and analyze it for bias.
Nina Cialone ’24
Cognitive Science Major and Political Science and Mass Communications Minors
AI Tools: ChatGPT, Botsonic, Custom GPT
I wrote for The Brown and White and applied for a freelance opportunity with Don’t Count Us Out Yet, a site and e-newsletter about Web 3.0 topics including the metaverse, artificial intelligence, the creator economy, genetic engineering, sustainable planet, and cryptocurrency. I began writing about ChatGPT but soon was writing about AI.
That’s when Craig Gordon ’76 asked me to take on a bigger challenge: create a chatbot for The Brown and White. It would allow reporters to search all of the paper’s old issues. I knew a bit of Java and had taken a Coursera class on programming, but I was far from a tech expert. I knew it would be an uphill battle but liked the challenge.
I began to research custom AI chatbot software. I was all in on one called BotSonic until I saw an ad for CustomGPT. After testing it, I knew it would serve students better. I then began the manual task of copying and pasting 6,733 URLs, every archived issue of The Brown and White, into the software. The chatbot can also draw answers from the full Brown and White website. Then testing began in order to discover any limitations and best practices.
The Brown and White staff now can talk to the chatbot and find out what has run in the paper about any topic. This project not only shows how someone like me, without a background in coding, can make AI work but also how journalism can adapt to new technologies.
Michael S. Lehman
Director, M.Eng. in Technical Entrepreneurship
Professor of Practice, Mechanical Engineering and Mechanics
AI Tools: ChatGPT and DALL·E
The faculty in the Technical Entrepreneurship (TE) program have purposefully integrated generative AI into courses focused on both product and venture. TE graduate students gravitate towards utilizing emerging technologies even if not yet perfected — they like learning on the cutting edge. Generative AI provides students with an additional viewpoint to consider, validate, debate, and include in their work as just one of many data points.
TE grad students look to ChatGPT to supplement their initial assumptions and research in their industry of focus (ranging from medical devices to fintech to sustainability): What is the wide range of stakeholder groups to interview, what are additional targeted discovery questions for those interviews, and what industry events should I consider attending? On the product-development side, ChatGPT provides an interesting scan of the myriad product categories in an industry. DALL·E can help visualize what their mock-ups will look like in their intended environment, providing context for the next prototype.
To date, the students and faculty in TE have been both pleasantly surprised and disappointed as they explore generative AI, prompting (pun intended) not only iterations but also robust classroom discussions and peer-to-peer learning.
Elizabeth Young-Miller
Social Sciences Librarian
AI Tools: ChatGPT, Keenious, Elicit, Dall·E
I was working with a professor in anthropology who asked students to create a visual ethnography of a subculture at Lehigh. We put that prompt in ChatGPT, and what it gave back was vague and had no images. It’s important for students to see this. AI at this point can generate some good prompts and keywords for general searches, but without clarity and specificity from the user or accurate and transparent sources from which to draw answers, the AI will hallucinate — make up things.