The Bootstrapped Founder

429: The Dead Internet Theory: Are We Building Machines That Only Talk to Other Machines?

December 26, 2025

Key Takeaways Copied to clipboard!

  • Founders must actively choose whether their AI tools contribute to the "dead internet" (machines talking only to machines) or augment genuine human connection. 
  • Generative AI should be used as a means (a guiding system or assistant for transformation) rather than the end (a pure generating system that replaces human creative work). 
  • Founders should implement AI to facilitate transformation and extraction/analysis, leaving the final creative or meaningful steps to the human user, rather than automating the entire process away. 

Segments

AI-Generated Social Media Observation
Copied to clipboard!
(00:00:00)
  • Key Takeaway: The speaker observed an entire LinkedIn thread composed solely of AI-generated posts and comments, highlighting the reality of the ‘dead internet theory’.
  • Summary: The speaker encountered a LinkedIn post and dozens of subsequent comments that were entirely AI-generated, representing a conversation with zero human involvement. This exchange, designed to look authentic, was described as both hilarious and tragic. This observation directly motivated the discussion on the dead internet theory and the role of founders in this trend.
Defining the Dead Internet Theory
Copied to clipboard!
(00:01:10)
  • Key Takeaway: The dead internet theory describes a network increasingly filled by AI systems communicating with other AI systems using AI-generated content.
  • Summary: This phenomenon is defined as machines creating content that pretends to be human, engaged with by other machines also pretending to be human, with LinkedIn being cited as the worst platform for this. This differs from useful machine-to-machine communication like APIs, which are designed for system interfacing. Founders using AI tools risk contributing to this network if their output is purely automated.
Sponsor Break: Paddle.com
Copied to clipboard!
(00:01:47)
  • Key Takeaway: Paddle.com acts as a merchant of record, handling taxes and currencies for software projects.
  • Summary: Paddle.com is used by the speaker as the merchant of record for all software projects. This service manages complex financial tasks like taxes and currency handling. This allows founders to focus solely on product development and user interaction.
Rethinking AI Tool Design
Copied to clipboard!
(00:02:53)
  • Key Takeaway: Founders should design AI tools to augment human connection rather than automate away the entire process, especially concerning content generation like cold emails.
  • Summary: A key risk is founders contributing to an internet where humans are left out of the communication loop. For tools like cold email generators, the focus should shift from full automation to providing a first draft or template that requires human review and personalization. The goal is to use AI as a guiding system, not a pure output generator.
AI as Means, Not End
Copied to clipboard!
(00:05:52)
  • Key Takeaway: Generative AI must be used as the means to facilitate transformation, acting as a guardrail, never as the final output that replaces human creative work.
  • Summary: This principle mirrors a version of Kant’s categorical imperative: interact with humans as the end, never just the means. Similarly, AI should facilitate transformation rather than completely automating it, ensuring it remains additive. The question for founders should be how AI can multiply or amplify the work being done, not replace the act of creative work.
PodScan: AI for Extraction, Not Generation
Copied to clipboard!
(00:07:13)
  • Key Takeaway: AI is best utilized in the background for tedious, non-creative data extraction and analysis, leaving subsequent decision-making to the customer.
  • Summary: In the speaker’s business, PodScan, AI extracts themes, topics, and names from transcripts but stops there, making the data available for human use. This avoids replacing the meaningful, creative work of interpreting that data. If a SaaS outcome is purely automated content, it actively contributes to the dead internet becoming reality.
Augmentative AI Use Case: List Expansion
Copied to clipboard!
(00:09:00)
  • Key Takeaway: AI can function as a background research agent to expand curated lists based on underlying data vectors, provided the final content creation remains human-driven.
  • Summary: The list similarity generation feature in PodScan uses AI to find related podcasts after the user has defined the initial vector of interest. This involves data acquisition and general-level value judgment, which AI handles effectively. The generated list is provided as raw material, not condensed into a finished article, keeping the human in the driver’s seat.
Conclusion: The Founder’s Choice
Copied to clipboard!
(00:11:08)
  • Key Takeaway: Founders face a choice: build tools that replace human thought and connection, or build tools that give humans superpowers while keeping them firmly in control.
  • Summary: The path forward involves deciding whether to add to the noise of machines talking to machines or to augment capabilities. The speaker advocates for the latter, where the machine acts as a guide, not the primary actor. This choice determines whether AI integration makes the internet less authentic or more powerful for users.
Call to Action and PodScan Promotion
Copied to clipboard!
(00:11:42)
  • Key Takeaway: PodScan monitors millions of podcasts for brand mentions and offers an AI agent identifying startup opportunities from podcast conversations.
  • Summary: The speaker directs listeners to find him on Twitter (@ArvidKahl) and promotes PodScan for monitoring brand mentions across over 4 million podcasts in real time. Additionally, ideas.podscan.fm uses an AI agent to identify startup opportunities based on ongoing podcast conversations.