---
title: "AI Cluetrain Manifesto: 95 theses for the internet after AI"
date: 2026-10-22T08:45:24Z
modified: 2026-10-22T08:48:52Z
permalink: "https://www.had.si/blog/en/2026/10/22/ai-cluetrain-manifesto-2/"
type: post
status: publish
excerpt: The AI Cluetrain Manifesto sets out 95 theses on artificial intelligence, trust, authorship and human conversation in an internet increasingly shaped by machines.
wpid: 390889
categories:
  - AI
tags:
  - AI
  - AI agents
  - AI Cluetrain Manifesto
  - AI slop
  - algorithms
  - Cluetrain Manifesto
  - generative artificial intelligence
  - SEO
  - english
  - pll_6ad9cd74e00d0
language:
  - english
post_translations:
  - pll_6ad9cd74e00d0
featured_image: "https://www.had.si/blog/wp-content/uploads/2026/10/ai_cluetrain_manifesto.jpg"
featured_image_alt: Ai Cluetrain Manifesto
timestamp: 2026-10-22T08:48:52Z
---

The AI Cluetrain Manifesto is a contemporary response to The Cluetrain Manifesto, published online in 1999 by Rick Levine, Christopher Locke, Doc Searls and David Weinberger. Its 95 theses began with a simple idea: networked markets are human conversations, not audiences, segments or corporate communication channels.

- [AI probably won’t take your job. Someone who knows how to use AI better than you might.](https://www.had.si/blog/wp-content/uploads/wp-mfa-exports/post/ai-probably-wont-take-your-job-someone-who-knows-how-to-use-ai-better-than-you-might.md)
- [Fact check / check the story about the French left-wing activist before sharing it](https://www.had.si/blog/wp-content/uploads/wp-mfa-exports/post/fact-check-french-left-wing-activist-story.md)
- [bloki.si, an AI tool for researching the property market / Thank you, Marko Plahuta](https://www.had.si/blog/wp-content/uploads/wp-mfa-exports/post/bloki-si-ai-tool-for-researching-the-property-market.md)

A quarter of a century later, the internet is changing again. Search is becoming an answer. People are talking to language models. Agents are beginning to communicate with agents. Content can be produced at industrial scale, while authenticity, attention, trust and human experience are becoming more valuable precisely because they are harder to produce.

These 95 new theses are not a replacement for the original manifesto, and the project is not affiliated with its authors. It is an attempt to pose a similarly uncomfortable question at a different technological moment: when machines enter the conversation, how do we make sure people do not disappear from it?

Markets are still conversations. The only question is who is really talking. The internet was meant to connect people. Somehow, we managed to connect bots to bots.

# AI Cluetrain Manifesto

1. The internet is becoming a conversation with machines.
2. Machines learned our language before we learned what to ask them.
3. AI does not understand the world. It understands our descriptions of the world.
4. A confident answer is not the same as a true answer.
5. Fluent speech is not intelligence.
6. Scale is not wisdom.
7. Prediction is not understanding.
8. Automation does not remove responsibility.
9. There is always a person somewhere behind the machine.
10. And usually someone else paying for the whole thing.
11. AI will create more words than humanity could ever read.
12. The problem is no longer creating information. The problem is deciding what deserves attention.
13. Attention is becoming more valuable than information.
14. When everything can be created, authenticity becomes scarce.
15. Scarcity creates value. The human voice will become scarce.
16. Perfect language is becoming suspicious language.
17. Typos may soon be proof of life.
18. The web once rewarded publishing. AI rewards being cited.
19. SEO wanted clicks. AI wants answers.
20. AEO, GEO and whatever acronym comes next will not fix bad content.
21. Writing for algorithms eventually makes everything sound the same.
22. When everyone optimises for machines, people lose interest.
23. The best optimisation may once again be having something to say.
24. You cannot create an original thought by stuffing in keywords.
25. AI can imitate a voice. It cannot live a life.
26. Experience remains stubbornly human.
27. A machine can describe heartbreak. It cannot miss anyone.
28. A machine can describe hunger without ever having been hungry.
29. That difference matters.
30. The web is filling up with content nobody wanted to write for people who never wanted to read it.
31. Publishing because you can is not the same as publishing because it is worth it.
32. Cheap content is becoming infinitely cheap.
33. An infinite amount of content creates a finite amount of patience.
34. AI slop is not a technological problem. It is an incentive problem.
35. Platforms demanded engagement. Machines gave it to them at industrial scale.
36. Be careful what you optimise for.
37. Metrics lose meaning when machines can produce them.
38. You can generate followers. You cannot generate trust.
39. You can buy traffic. You cannot buy credibility.
40. Reach is not a relationship.
41. Engagement is not affection.
42. Virality is not significance.
43. Popularity is not truth.
44. A consensus created by machines is still a manufactured consensus.
45. The dead internet theory was more entertaining before companies started making it real.
46. Bots talking to bots are not a community.
47. An AI agent buying from another AI agent is still spending someone’s money.
48. Delegation requires trust. Autonomy requires even more.
49. Convenience always sends someone the bill.
50. If AI decides for you, you should know whose interests shaped that decision.
51. Recommendation systems are opinions dressed up as maths.
52. Algorithms are not neutral just because they contain numbers.
53. Training data contains history, prejudice, genius, stupidity and us.
54. AI inherited the internet. The internet should be worried about that.
55. Garbage in. Confidence out.
56. Being cited is becoming a form of power.
57. Absence from machine knowledge will soon resemble absence from the web.
58. If machines mediate knowledge, whoever shapes the machines also shapes reality.
59. We once feared gatekeepers. Now the gatekeeper answers us.
60. A search engine showed us ten links. AI gives us one sentence.
61. That sentence has enormous power.
62. Sources matter most precisely when users see them least.
63. Transparency is not a footnote.
64. AI said so is not an argument.
65. The algorithm decided is not an argument either.
66. People built these systems. People can question them too.
67. The right to ask why becomes more important as automation grows.
68. The right to say no matters too.
69. Not everything needs AI.
70. If you add AI to a bad product, you get a bad product with AI.
71. A chatbot is not a strategy.
72. A button labelled AI is not innovation.
73. If we call everything AI, the term will eventually mean nothing.
74. Artificial intelligence has created a surprising amount of artificial marketing.
75. Companies that say AI-first should explain the problem first.
76. Technology should solve problems, not justify pitch decks.
77. The future does not need another AI wrapper.
78. Sometimes a database is still just a database.
79. Sometimes one if statement is enough.
80. Sometimes a person is a better API.
81. AI should remove tedious work, not meaningful work.
82. Productivity is not measured by the number of words produced.
83. Speed does not improve a bad decision. It only delivers it sooner.
84. Automating stupidity produces stupidity faster.
85. Efficiency without purpose is merely acceleration.
86. We will use AI to become more human or less human. That choice remains ours.
87. The most valuable skill may be knowing when not to ask the machine.
88. Thinking is slower than generating. Think anyway.
89. Doubt is a feature, not a bug.
90. Curiosity must survive convenience.
91. Creativity begins where autocomplete ends.
92. The internet of the future will contain more machines. It must not contain fewer people.
93. We do not need machines that sound more human. We need people to stop sounding like machines.
94. Markets are still conversations. The only question is who is really talking.
95. The internet is ours only for as long as we act as though it truly is.

This is not anti-AI nostalgia. It is not complaining about the future, and it is not a marketing stunt. It is a human-scale manifesto about language, trust, authorship, attention and responsibility in a web increasingly shaped by machine-generated speech.

Source: [https://www.had.si/cluetrain/](https://www.had.si/cluetrain/)