Whitman College’s library, Penrose, took a chance on AI last semester. Beta-testing a new generative AI-powered research tool called WOS (Web of Science), it seemed the library was embracing what many have referred to as the AI revolution. 

Functioning like a ‘digital brain,’ the WOS Research Assistant uses generative AI to simplify literature searches through natural language queries. 

Josep Meindertsma, founder of PauseAI, an organization working to halt global AI expansion to avoid human extinction, explained the distinction between AI and its generative counterpart. 

“GenAI [generative AI] is different from many of the earlier forms of AI. For a long time, ‘AI’ consisted mostly of software written by programmers. With modern neural networks powering GenAI, that changed completely: Now, AI developers only write a training algorithm, spin up a digital brain, and feed it with terabytes [units of digital data measurement equivalent to one trillion bytes] of data,” Meindertsma said in an email to the Wire. 

Generative AI offers an alternative to traditional Boolean search methods, which use operators (like “and,” “or” and “not”) and modifiers (parentheses, asterisks and quotation marks) to refine searches. During the October trial period, Whitman faculty and students explored the tool’s features such as topic suggestions, data visualization and streamlined searches across scientific literature.  

However, despite the library’s promotion of the tool which touted its “insights, visualizations and prompts,” its ability to “tackl[e] complex research tasks,” and initial public interest, staff ultimately decided not to implement the service. 

“The Whitman College Penrose Library does not have an active subscription to this tool nor do we currently intend on purchasing it,” said Lindsay Tebeck, Head of Collection Strategies of Penrose Library. “ …the feedback that we received from Whitman community members informed our decision.”  

Concerns arose regarding the tool’s search execution and pricing. Tebeck explained that while some users appreciated the assistant’s AI-generated insights, others found its search methods less transparent and less selective compared to traditional Boolean queries.  

“The execution of the AI prompts is more enigmatic than the traditional Boolean search,” Tebeck said. “The traditional Boolean search gives the user more selective power [through the use of keywords and logic statements].” 

Additionally, the WOS Research Assistant’s reliance on PageRank-style algorithms, which hierarchize websites based on their relative importance, posed issues with bias, potentially prioritizing highly cited papers while overlooking emerging research.

The library hasn’t completely cast out AI, however. In fact, its website features the Primo Research Assistant, a generative-AI apparatus similar to the WOS tool in terms of source provision and synthesis. 

Subscription or not, the discussion surrounding AI tools like the WOS Research Assistant reflects broader conversations at Whitman about AI’s place — or lack thereof — in education. 

Professor of Mathematics Albert Schueller integrates discussions about AI into his courses.  

“We spend time on the first day of class talking about what generative AI is and how it can be used appropriately,” Schueller said. “Students discuss how they’ve used AI in the past and what they think would be a reasonable way to use it in the class.”  

Schueller encourages AI use for learning and problem-solving, but not as a shortcut. 

“Using AI to get help when you’re stuck is appropriate — tell the AI where you are in your thought process and ask for a hint, but don’t let it solve the problem for you,” Schueller said. “Friction is where learning happens, and there are no shortcuts for that.”  

He also pointed out that AI cannot replace deep content knowledge. 

“If you don’t understand calculus, AI won’t do the thinking for you,” Schueller said. “To be productive with AI, you still need a strong foundation in your subject.”  

Meindertsma acknowledges the utilities of AI in certain situations. 

“I personally think AI models can be very useful in the right contexts — I use them all the time as a software engineer. Still, I believe we should be extremely careful when these systems start to gain dangerous capabilities (like cybersecurity skills), which we’re now getting awfully close to,” Meindertsma said. 

While Penrose will not be adopting the WOS Research Assistant at this time, staff say they remain open to exploring AI-driven tools. 

“We value community input and will continue to evaluate new tools as they emerge,” said Tebeck.  

Meanwhile, Schueller believes that AI’s role in education will continue to evolve but stresses that human interaction remains irreplaceable. 

“There’s something valuable about being in the same room, talking to each other,” Schueller said. “AI can assist, but it can’t replace the learning experience.”  

For now, Whitman students and faculty are encouraged to stay informed and engage in discussions about AI’s role in research and learning. 

Tebeck stressed the importance of being able to identify misinformation in the age of AI. 

“I encourage users to practice checking the accuracy and credibility of any given source, AI or not. As AI tools continue to grow in popularity, information literacy is an essential skill for researchers of all experience levels,” Tebeck said. 

At Whitman, it can’t hurt students to ask their professors about their AI policies. 

“Have a conversation with your professor,” Schueller said. “Understand what’s appropriate and how AI fits into your education — because whether we like it or not, AI is here to stay.”