/
RSS Feed
Show Notes: AI in the Workplace: Helpful Tool or Overhyped Magic?
Podcast: Nerding Out with That Nerdy Catholic Hosts: Seth Paine (Programming/Video Production background) & Cesar Rios (Electrical Engineering background) Guest: Lee Sondeno, Director of Young Adult Ministry at St. Victoria (Victoria, MN) and 18-year IT industry veteran
Episode Overview
In this episode, the team ditches the purely theological speculation to look at Artificial Intelligence through the lens of practical, real-world tech experience. Combining backgrounds in programming, engineering, and IT, Seth, Cesar, and Lee dissect the current corporate hype surrounding AI, share cautionary tales of AI errors in the wild, and discuss the critical differences between using AI as an automation tool versus blindly trusting it as a “person in a box.”
Key Discussion Points
- The “Test but Verify” Rule & AI Hallucinations
- The Vatican AI Policy Blunder: Lee shares a firsthand account of a company that tried to draft an official AI policy. To ensure it was aligned properly, executives tasked an AI agent with drafting the policy specifically based on official Vatican documents.
- Two Layers of Fabrications: While the document initially looked professional and included realistic-looking sources and URLs, a deep dive revealed the AI had completely hallucinated. The links went nowhere, and the specific “quotes” attributed to the Vatican didn’t actually exist anywhere in the real documents.
- Even the Creators Agree: Lee shares a striking quote from OpenAI’s Sam Altman admitting that ChatGPT should be the technology you trust the least right now because of its severe tendency to hallucinate.
- The Illusion of the “Magic Box” vs. Actual Efficiency
- The Coworker with Bravado: Seth compares AI to a overly confident coworker who acts like they know everything, easily gains unearned responsibilities, but is ultimately just fabricating answers to move up.
- Workplace Realities: Despite Wall Street Journal reports highlighting that AI is often increasing—rather than lightening—workplace intensity and density, many companies have blind faith in the tech.
- The Over-Complication of Code: Seth and Cesar share an experiment where they asked an AI model to write a snippet of code. While human programmers could write the clean script in 3 to 5 lines, the AI generated 30 bloated lines to accomplish the same basic task. The time required to repeatedly prompt, check, and debug the bloated output often doubles or triples the work time.
- The Tech Layoff Hype & Corporate Bubbles
- Blaming AI for “Right-Sizing”: The guys discuss recent corporate announcements where tech giants like Meta laid off thousands of employees, explicitly claiming AI was replacing the workforce.
- The Shareholder Game: Lee points out that tech companies like Meta doubled their workforce between 2020 and 2022 (e.g., jumping from 40k to 80k employees). The current wave of layoffs is actually a natural corrections of the pandemic-era IT bubble, but executives lean heavily into the “AI replacement” narrative because it drives shareholder excitement and raises stock values.
- The “Follow the Leader” Mentality: Cesar explains how non-tech legacy industries (like glass container manufacturing) are being led by financial or commercial executives who blindly copy tech giants without understanding how the underlying technology actually works.
- How to Properly Use AI: A Tool, Not a Replacement
- Historical Context (The Mechanical Turk): Lee references the famous 18th-century “chess-playing automation” machine, which amazed crowds who thought it was magic, when in reality, it was just a skilled human chess player hidden inside a box. Similarly, modern AI is often just repackaging and regurgitating human labor.
- The Dreamweaver vs. FrontPage Analogy: Seth maps the current AI landscape back to 1990s web design. Tools like Adobe Dreamweaver gave programmers clean, adjustable foundational visual layouts , whereas Microsoft FrontPage spat out heavily bloated, messy nested code that required endless human tweaking.
- Sophisticated Prompting Requires Human Experience: The team concludes that AI can be a spectacular automation tool—such as speeding up visual layouts or generating basic video descriptions—but it only functions well if the human using it has the knowledge and experience to ask the right questions and rigidly cross-check the final output.
Notable Quotes
- “It should be the tech that you don’t trust that much. AI hallucinates.” — Sam Altman (quoted by Lee Sondeno)
- “It feels like a coworker that has such bravado and such confidence… and then at some point you say, ‘Wait a second, this guy has been making all this stuff up all along.’” — Seth Paine
- “It’s not that AI can do the work, it’s that it is a tool and that’s the thing that nobody seems to understand. It’s a very good tool. It’s not a very good person.” — Cesar Rios
Links & Resources Mentioned
- Check out our main hub: thatnerdicatholic.com
- Grab some Nerdy Catholic swag (t-shirts, mugs, prints): thatnerdicatholic.com/merch
Leave a Reply