Podcast Summary

Podcast: Science Vs
Episode: Is AI Making Us Stupid?
Host: Meryl Horn (filling in for Wendy Zuckerman)
Producer: Spotify Studios
Date: December 18, 2025

Episode Overview

This episode examines the provocative question: Is AI destroying our ability to think? Host Meryl Horn and senior producer Rose Rimler dive into the scientific evidence behind fears that AI tools like ChatGPT may be making us “stupider”—or whether they might actually boost our productivity and intelligence. The episode features scientific studies, expert opinions, and lively discussion about the impact of AI on our brains, our work, and the future of learning.

Key Discussion Points and Insights

1. The Panic and Promise Around AI Usage

• Concerns: AI use is skyrocketing, with surveys showing 62% of US adults and 80% of students interacting with AI for various tasks ([00:29]).
• Public anxiety centers on the potential for AI (especially tools like ChatGPT) to erode memory, learning, and creativity.
• Flip side: Tech enthusiasts and some scientists claim AI enhances productivity, speeds up busy work, and even accelerates scientific progress ([01:51]).

2. Does AI Make Us Stupider? Reviewing the Evidence

Case Study: Learning with ChatGPT vs. Google Search

• Lead researcher: Shiri Melumad, University of Pennsylvania ([06:22])
• Study design: Over 10,000 adults were given research tasks (e.g. “How to plant a vegetable garden?”) using either ChatGPT or standard Google search.
• Findings:
  • ChatGPT group: Advice was sparser, more generic, and referenced fewer facts ([07:58], [08:26]).
  • Google group: Responses were richer and carried more personal, creative language—even if not always factually perfect ([09:39]).
• Notable quote (Melumad):

  “It’s unique to the writer. It really doesn’t come off as generic as the ChatGPT pieces.” ([11:09])

• Participants using ChatGPT:
  • Felt they learned less ([11:34])
  • Produced advice that was rated less helpful by outside observers ([12:13])
• The issue is not just feeling less engaged, but actually learning less ([12:48])
  • "These syntheses that LLMs provide are transforming learning from a more active to a more passive process. And that’s what we’re losing." —Shiri Melumad [13:16]

Your Brain on ChatGPT

• Small brainwave (EEG) study ("Your brain on ChatGPT") found weakest brain region connectivity during ChatGPT use versus Google use or unaided thinking ([14:18]).
• Suggests lower mental engagement when using AI for cognitive tasks.

Memory and De-skilling

• Studies suggest people remember less of what they wrote or learned if they used ChatGPT ([15:18]).
• Real-world concern: Example of doctors using AI for colonoscopies, then performing worse when AI was removed—potential evidence for "deskilling" ([18:47]).
  • “We’ll just lose these abilities thanks to AI.” — Meryl Horn [18:09]

3. The Pros: Can AI Make Us Smarter or More Productive? ([22:21])

AI as an Enhancer, Not Just a Crutch

• Guest: Aaron French, Kennesaw State University ([23:00])
• Perspective: Outcomes depend on how AI is used. It can be a crutch that slows learning or an enhancer that enables us to do more:

  “For some people, it’s absolutely going to make them, I guess I don't want to say dumber, but they're not going to learn or improve because of that. Other people… are going to be able to do more with AI than they were able to do without it.” —Aaron French ([23:00])

Time Savings Across Professions

• Studies show AI saves significant time for dietitians, programmers, teachers, researchers.
  • Example: Teachers saving an average of 6 hours a week ([24:47]).
• Key question: Do people invest saved time in meaningful or creative work, or is it wasted on distractions?
  • Early studies suggest ~36% of managers admit to “wasting” at least half their saved time ([27:00]).

Historical Parallel: The Calculator Debate

• When calculators first entered schools, similar worries existed (“calcuholics”) ([28:43]).
• Meta-analysis of 50+ studies: Calculator use did not harm basic math skills; in fact, it improved higher-order problem solving and attitudes towards math ([30:07]).
• Calculators enabled more complex learning; AI may do the same.

AI as a Tool For Learning

• Studies show potential for AI as a tutor or assistant for students with specific needs ([31:59]).
• Cautions remain about overreliance and the risk of teachers being replaced in meaningful feedback ([32:10]).

AI Driving Scientific Progress

• AI's successes in fields like predicting protein structures, finding black holes, and accelerating data analysis ([33:11]).
• Over 50% of 1600 surveyed scientists see AI as “very important or essential” to their fields over the next decade ([33:11]).

4. Balancing Act: Avoiding “Homogenization” and Loss of Creativity

• Major concern is homogenization: AI-generated responses tend to be less creative and more generic ([34:07]).
• Meryl Horn aims to avoid AI for creative tasks to preserve originality.
• Even researchers who found negative effects still use AI for minor, non-essential queries ([34:50]):

  “To the extent that you actually care about learning more deeply about something, you should really try to avoid starting off your research with an LLM because it's too tempting to stop with the syntheses that you're provided.” —Shiri Melumad ([34:50])

5. Outlook: Will the Next Generation Be Less Skilled?

• No conclusive long-term data yet on AI's impact on students, as tools are too new for robust longitudinal studies ([17:19], [35:20]).
• Meryl and Rose reflect on similar past panics over the Internet and calculators, but note humans have adapted so far:

  “I’m not that worried yet. ...Were we really all totally stupid when we were cavemen? I don't think so. I think we were probably okay for whatever that context required of us.” —Meryl Horn ([35:39])

Notable Quotes & Memorable Moments

• “Cheating is the new major.” ([01:14], speaker B)
• “Every single time we give it a prompt, our own brain cells are burning.” ([00:29], speaker A)
• “The advice kind of sucks when people use large language models, whether it's ChatGPT or Google's AI overview.” ([12:13], Meryl Horn)
• "When I start with ChatGPT to learn about something, I find it really hard to...keep learning more.” ([34:50], Shiri Melumad)
• “Is using AI the equivalent of shoving your brain into the microwave?” ([01:52], speaker A)

Timestamps for Important Segments

• [00:00] Episode intro, stating the main question/debate
• [03:48] Personal experiences: How the hosts use ChatGPT
• [06:22] Introduction of Shiri Melumad’s study on learning with AI vs. Google
• [08:53] Listening to examples of ChatGPT vs. Google-generated advice
• [11:34] How participants felt about their advice/learning
• [14:01] “Your Brain on ChatGPT” EEG study reviewed
• [15:18] Studies on memory decline when using AI
• [18:47] Medical study on de-skilling among doctors using AI
• [22:21] Segment on productivity and the potential positives of AI
• [23:00] Interview with Aaron French—AI as an enhancer or a crutch
• [24:47] AI saving teachers hours per week
• [28:43] Calculator "calcuholics" parallels and meta-analysis findings
• [31:59] AI as assistive tech and customizable tutor
• [33:11] AI's role in cutting-edge science
• [34:50] Advice on when to avoid using AI for creative, deeper learning
• [35:20] Discussion about generational skill loss fears and historical context

Summary Takeaways

• AI can both harm and enhance learning and productivity, depending on how it is used.
• Using AI for rote tasks can free up time, but that time may or may not be used productively.
• Evidence suggests overuse of AI for learning (especially for students) can diminish engagement, memory, and creativity.
• However, AI’s positive impact is clear in science, when used thoughtfully.
• Lesson: Use AI as a tool for efficiency and accessibility, but avoid it for deeper, creative, or meaningful learning if you want to preserve and build those skills.

Science Vs signs off reminding listeners that the AI question, much like past debates about calculators or the Internet, is really about how humans adjust and adapt to new tools—and where we draw the line between convenience and cognitive engagement.