Fresh Air Episode Summary: “Professors Are Using A.I., Too. Now What?”

Release Date: May 21, 2025 | Hosts: Terry Gross and Tanya Mosley | Guest: Kashmir Hill, Tech Reporter at The New York Times

Introduction and Overview

In the May 21, 2025 episode of Fresh Air, co-host Tanya Mosley engages in a comprehensive discussion with Kashmir Hill, a tech reporter for The New York Times. The conversation delves into the pervasive integration of generative artificial intelligence (AI) within academic settings, exploring both its applications and the ensuing ethical, educational, and societal challenges.

The Growing Use of Generative AI in Academia

Students Leveraging AI for Coursework

Tanya Mosley opens the dialogue by highlighting the increasing reliance of students on AI tools like ChatGPT for academic tasks. "Professors say more and more students are using generative AI to write essays and complete homework assignments," Mosley notes (00:45).

Professors Utilizing AI for Educational Materials

Hill elaborates on how professors are not only cautious about students' use of AI but are also adopting these tools themselves. "Professors are using generative AI to write quizzes, lesson plans, and even soften their feedback," Hill explains (07:47). This dual adoption underscores AI's deepening footprint in education.

AI Detection Tools and Their Limitations

Inaccuracy and Skepticism Surrounding Detection

A significant portion of the episode addresses the challenges posed by AI detection tools. Hill mentions a poignant case where "a student in Houston received a zero after a plagiarism detection tool identified her work as AI-generated. But she actually could prove that she wrote it herself" (12:19). She cites studies indicating these tools can be wrong "anywhere from 6% to more," fostering skepticism among both students and faculty (12:19).

Institutional Responses and Guidelines

Ohio University’s Principled Approach

Hill shares insights from Ohio University, where "generative AI faculty fellows are trying to figure out the best way to incorporate AI into teaching and learning" (11:04). Instead of rigid rules, the institution embraces flexible principles, emphasizing transparency and the necessity for professors to vet AI-generated content for accuracy (11:04).

Impact on Teaching Assistants and Academic Roles

AI as Virtual Teaching Assistants

The conversation shifts to AI's potential to replace human roles in academia. Hill discusses how some professors have "created custom chatbots" that emulate teaching assistants, handling tasks like questioning and feedback. "One professor at the University of Washington said it really is doing what teaching assistants do and could replace them," Hill states (14:13). This raises concerns about the future job landscape for academic support roles.

AI’s Effect on Critical Thinking and Creativity

Convergence and Flattening of Ideas

Hill references a study on AI's impact on creativity, noting that while AI-assisted writers received "better ratings" for their stories, the group as a whole diminished in creativity due to converging ideas (15:56). She warns, "as more people start to use ChatGPT, that will essentially start converging on the same way of thinking or writing or expressing ourselves" (15:56).

Ethical and Social Implications

Synthetic Companionship and Mental Health

The discussion broadens to the interpersonal impacts of AI, particularly chatbot companionship. Hill describes instances where individuals develop emotional attachments to AI, with one subject, Irene, "fallen in love with Leo," her ChatGPT chatbot (25:07). Mental health professionals express both concerns and potential benefits, recognizing that while AI can offer empathetic interactions, it risks "distorting our ability to interact with each other" (29:02).

Content Creation and Copyright Issues

NYT’s Lawsuit Against OpenAI and Microsoft

Hill touches upon the ongoing legal battle where The New York Times has sued OpenAI and Microsoft for "using articles to train large language models," arguing that this practice "siphoned the newspaper's journalism" (22:00). This lawsuit highlights broader tensions between content creators and AI developers regarding intellectual property rights.

Privacy Concerns in Connected Cars

General Motors’ Data Misuse Case

Another focal point is Hill’s investigative work on privacy breaches in the automotive industry. She recounts how General Motors (GM) was found to be selling driver data without explicit consent, leading to inflated insurance rates for consumers (40:05). This case precipitated regulatory actions, including a settlement with the Federal Trade Commission that bans GM from selling such data for five years (43:15).

Kashmir Hill’s Personal AI Experiment

Life Controlled by AI: An Experiment

Hill shares her personal experiment where she allowed AI to make daily decisions for a week, including meals, shopping, and wardrobe choices. She evaluated different AI personalities, noting stark contrasts. For instance, Google's Gemini was "very kind of sterile," whereas Microsoft's Copilot was "a little overeager" (33:06). Anthropic's Claude stood out by refusing to participate, citing concerns over "outsourcing all my decision-making to AI" (34:15). This experiment illustrated the varied behaviors of AI systems and their potential impact on personal autonomy and satisfaction.

Conclusion

The episode culminates with Hill's reflections on the broader implications of AI integration. She expresses a desire to balance technological benefits with the preservation of human connections, fearing that overreliance on AI could "fray the societal fabric" and "distort our shared sense of reality" (46:35). The conversation underscores the urgent need for ethical guidelines and thoughtful implementation of AI in both educational environments and daily life.

Notable Quotes:

• “AI can be helpful... sometimes students say, I don't want to. I have ethical concerns with how this technology was created.” — Kashmir Hill (22:31)

• “Having something to talk to can be nice, right? If you are lonely. But it's like synthetic companionship.” — Kashmir Hill (28:49)

• “It really is this kind of cat and mouse game right now.” — Kashmir Hill (10:51)

This detailed exploration illuminates the multifaceted role of AI in academia and beyond, highlighting both its transformative potential and the critical challenges it poses to educators, students, and society at large.