Big Technology Podcast (Sept 19, 2025): "How People Use ChatGPT, Meta’s New AI Glasses, Can Jimmy Kimmel Be Canceled?"

Main Theme

This Friday edition of the Big Technology Podcast, hosted by Alex Kantrowitz with regular guest Ranjan Roy (Margins), dives into fresh data on how people are using generative AI chatbots like ChatGPT and Claude, scrutinizes Meta’s new AI Ray-Ban glasses (including their troubled live demo), and discusses the meaning of “cancel culture” in 2025—particularly in the case of Jimmy Kimmel’s show being paused by political pressure.

Key Discussion Points & Insights

1. How People Use ChatGPT: OpenAI’s New Data Drop

[02:40–29:10]

• OpenAI’s User Data:

  • Three-quarters of ChatGPT conversations are for practical guidance, seeking information, and writing.
  • Writing is the most common work-related use, while coding and self-expression are niche.
  • Everyday tasks (tutoring, teaching, creative ideation) dominate usage.
  • Companion-style interactions (relationships, roleplay) are rare (~1.9% of messages).

  "Practical guidance, seeking information and writing are the three most common topics and collectively account for nearly 80% of all conversations."
  — Alex Kantrowitz [03:00]

• Is ChatGPT Redefining Search?

  • Debate on whether AI is replacing or supplementing search: Ranjan notes practical guidance and information-seeking are core to traditional search, while Alex emphasizes AI is a technological leap (“layer”) beyond merely reshuffling existing web content.
  • Nods to the old SEO-driven “how-to” web (eHow, Howcast) versus AI’s instantly generated on-demand advice.

• OpenAI’s Usage Categories & Ranjan’s Critique:

  • OpenAI splits messages into “asking” (49%), “doing” (40%), and “expressing” (11%).
  • Ranjan argues OpenAI’s definition of “doing” (help drafting text, planning) is too broad and not truly "agentic": The bot stops short of real-world action, such as sending an email.

  "They define doing as getting help in doing some kind of process, like drafting text... To me, doing is actually going out and doing it."
  — Ranjan Roy [08:45]

• OpenAI’s Framing: Marketing or Neutral Research?

  • Both hosts express skepticism about corporate research spun as “objective.”
  • The PDF’s “academic” look lends credibility, but findings, hosts say, conveniently fit and push OpenAI’s agentic business narrative.
  • Demographics: Usage is getting younger (half aged 18–25), and gender gaps are closing.
  • Only 1.9% of messages involve relationships, 0.4% roleplay—yet that still means millions of companion-style chats per week with 700M+ users.

• The Trust Factor & Economic Impact:

  • Users increasingly lean on chatbots for decision support and personal guidance, not just factual queries.
  • Ranjan and Alex discuss how “trust” in AI bots is already high—people rely on them for health advice, party invitation writing, and more.

  "Just the amount of trust people have in these bots is unbelievable already."
  — Alex Kantrowitz [22:50]

  • Decision support might be showing up in productivity, but its economic impact is hard to measure in GDP-style stats.

• Is the Web in Decline?

  • People may prefer AI answers because the open web became overrun by SEO spam and unreliable content.
  • "Is the web dead?" A recurring debate on the podcast.

• ChatGPT’s Explosive Growth:

  • Usage plateaued in early 2024 but then rocketed—projected to 800M–900M weekly users by end of 2025.

  "That is astonishing growth. And I have never seen this in any product ever before."
  — Alex Kantrowitz [28:30]

2. Anthropic’s Data: How CLAUDE is Used in Business

[30:23–40:19]

• Enterprise Adoption:

  • 40% of US employers use AI at work—double from 2023.
  • AI is spreading far faster than historic technologies (electricity, PCs, Internet).

  "Even the rapidly adopted Internet took around five years to hit adoption rate that AI has reached in just two years."
  — Alex Kantrowitz [31:30]

• Claude’s Usage Patterns:

  • Coding dominates (36% of business use), but educational and scientific queries are rising.
  • “Directive” conversations—where users ask Claude to complete tasks autonomously—are up noticeably.
  • Anthropic claims that automation (rather than augmentation) is now the top use, but hosts question the fuzziness of these definitions.

  "Are people trying to augment or automate work? The study seems to suggest that automation is taking priority."
  — Alex Kantrowitz [38:00]

• Hosts’ Nuanced Take:

  • Both caution that usage stats often neatly fit company marketing objectives.
  • Noting the odd nugget that Washington, DC per capita underpins Claude usage—implying heavy government adoption.

  "When all these numbers just so neatly fit into the existing strategic narrative of these companies, I have a hard time just, you know, taking it at full face value."
  — Ranjan Roy [39:51]

3. Meta’s New $799 AI Glasses

[40:38–47:33]

• Product Overview:

  • New Ray-Ban-style glasses with a display, live captions, music apps, and messaging (“offloading” tasks from phone to eyewear).
  • Demo failed at Meta event (due to “DDoS-ing” all servers in the room; all devices triggered at once).

• Divergent Reactions:

  • Reviewers onsite loved the glasses, but demo failure highlighted live tech’s risks.
  • Ranjan is bullish: sees potential in “lightly augmented” computing; data at the periphery could keep people off their phones.

  "This is about keeping your phone in your pocket. That's what, that's exactly how I've seen this whole interface of computing for a long time."
  — Ranjan Roy [44:10]

  • Alex is skeptical: worries about even deeper digital immersion, potential social awkwardness (“date” parodies, always-on prompts, etc.).

• Social Dynamics & Future:

  • Both anticipate new social normalcies (“Are you scrolling Reels while I’m talking to you?”).
  • History of wearable tech (Google Glass parodies, Snap Spectacles) hints at both opportunity and awkwardness.

4. Can You Still ‘Cancel’ a Comedian in 2025? The Jimmy Kimmel Case

[47:33–54:26]

• Background:
  Jimmy Kimmel’s late-night show is paused after political pressure—raising questions about whether “cancel culture” works as it once did.

  "If you're going to be taken seriously, you have to be willing to, you know, to be confident enough to take ridicule. Goes for the left and the right."
  — Alex Kantrowitz [48:00]

• Changing Media Power Structure:

  • The old model (major networks, centralized shows) is less relevant—today, audience ownership is everything.
  • Hosts cite examples: Shane Gillis, Tucker Carlson, Meghan Kelly, Bill Maher, Conan O’Brien, and even Jeffrey Jarvis's viral CNN moment (“the only good news I see...is that mass media are dying”).

  "You can get canceled. And if, especially if you make that a central part of your they're coming for me in the mainstream media... It's become such an almost like frustrating playbook, but a successful one."
  — Ranjan Roy [50:10]

• Modern "Cancellation" = Going Independent:

  • Kimmel or Colbert could leave and still thrive by owning their own audiences (YouTube, podcasts, Substack, etc.).
  • The next generation of comedians will see direct digital connection as default, not network TV as pinnacle.

  "The government's power to do this type of stuff is going to be much, much more limited because they don't have the ability to like go to a YouTube and say yank Jimmy Kimmel."
  — Alex Kantrowitz [54:20]

Notable Quotes and Memorable Moments

• On OpenAI’s stats:
  "The greatest form of content marketing you can do is make it look like a research paper and then suddenly just add so much credibility to it."
  — Ranjan Roy [13:25]

• On Companion Use-Case Reality:
  "This would say 1.9% of people have some sort of relationship with them. That would still leave us with 13.3 million people weekly having a relationship style conversation with ChatGPT. Obviously it's not the overriding use case but that's still a shit ton of people."
  — Alex Kantrowitz [16:24]

• On the Future of 'Cancellation':
  "If Kimmel wanted to, he could go all in blue sky, subscribe to my substack and like, be just as influential."
  — Ranjan Roy [50:45]

Timestamps for Important Segments

• [02:40] OpenAI ChatGPT usage data discussion begins
• [09:23] Thought partner vs. agentic uses, and critique of OpenAI’s categories
• [13:14] How companies frame research for marketing impact
• [22:50] Trust in chatbots and economic implications
• [28:23] Explosive growth numbers for ChatGPT
• [30:23] Anthropic/Claude usage report begins
• [38:00] Automation overtaking augmentation in Claude usage
• [40:38] Meta’s AI glasses, demo drama, and debate
• [47:33] Jimmy Kimmel situation: Can cancellation still “work”?
• [54:26] Summary thoughts on the changing nature of media gatekeeping

Tone

Conversational, skeptical (especially toward corporate PR), and full of tech-world in-jokes and cultural references. Both hosts clearly combine industry-insider knowledge with a playful but critical approach.

Summary

This episode is a deep dive into how generative AI is actually being used, the narratives tech companies build around their products, the rapidly evolving landscape of consumer AI hardware, and the greatly diminished (or transformed) power of “cancellation” in a fragmenting media world. Kantrowitz and Roy emphasize the need for skepticism, context, and nuanced interpretation of all tech narratives—especially those served up as “research.”

If you want to understand not just the numbers, but the meaning behind the numbers in AI adoption, and how tech is reshaping (and being spun in) everyday life and media, this episode is rich with insights.