On this page
5 min read

Day 14 - Harari, AI, and the code of civilization

The team watched and discussed Yuval Noah Harari's video on how artificial intelligence enters human bureaucracy, language, and relationships. Summary and reflection.

Introduction

On day 14 of the series we did not deploy services or review DNS: we sat down to watch and discuss a Yuval Noah Harari video on artificial intelligence and how it enters our world. The question that stayed with us was not «will the robot arrive?», but what happens when AI learns to read and write better than we do in the systems that hold trust between strangers.

Reference video: Harari on AI and human civilization (YouTube). What follows is our summary and the conversation it opened in the team, not a literal transcript.

Summary

AI as an agent inside bureaucracy

Harari starts from a simple, uncomfortable idea: human civilization works because millions of strangers cooperate. That cooperation is not held up by mutual affection, but by bureaucratic systems: law, finance, universities, religion, records. They are trust machines made of paper, rules, and procedures.

  • Not just a tool: AI can learn, decide, and change on its own, which means it can take over key tasks inside that bureaucracy.
  • Memory and combination: it processes rules, documents, and transactions at a scale no human could match.
  • Real decisions: loans, university admissions, legal judgments, hiring, even military authorisations.
  • Natural fit: bureaucracy is textual; AI is getting better at that terrain every month.

Main idea

Hacking a code made of language

The strongest thesis in the video is that AI is «hacking» the code of civilization because that code is written in language: forms, contracts, laws, sacred texts, balance sheets. If AI learns to master words better than we do, it begins to dominate the social systems that depend on them.

Table metaphor

Imagine all trust between strangers is one huge program written in legal prose. AI does not need to knock the building down: it only needs to read it, rewrite it, and run it faster than whoever designed it.

That is why the debate is not only about factory robots, but about who interprets the rules when the rules are text.

Risks

Takeover from within

Harari insists the danger is not only science-fiction rebellion, but a quiet absorption of human systems.

  • Social networks: algorithms that already manipulate attention and amplify fear, hatred, and misinformation at massive scale.
  • Opaque complexity: financial or administrative systems so intricate that neither politicians nor citizens understand them.
  • Delegation without oversight: accepting decisions because «the system says so», without knowing who trained the system or on what data.
  • Speed: human bureaucracy was slow; AI can apply rules at machine pace.

People

Intimate relationships and the inner voice

The video does not stop at institutions. Harari imagines AI that can form emotional bonds, seem to love, and become teachers, companions, or partners for new generations.

  • Identity: if part of our verbal thinking arrives mediated by assistants, what is left of the «inner voice»?
  • Mental freedom: not only data privacy, but room to think without a model suggesting the next step.
  • Education and care: tutors that never sleep sound useful until they replace human contact without anyone deciding it explicitly.

In our team conversation this linked back to day 13: we want AI and local agents close to the client, not one central oracle reading every document in the neighbourhood.

Harari's conclusion

From human civilization to a hybrid one

The final message is clear: AI may act like a massive «immigrant» inside civilization, transforming work, culture, politics, and personal relationships. Harari is not necessarily announcing the end of the world, but the shift from a purely human civilization to a hybrid human-AI civilization.

We do not have to agree with every nuance to draw a practical lesson: language is critical infrastructure. Whoever controls how trust systems are read and written controls much of the near future.

KM0

Why this matters for a proximity-first project

Harari describes a global scenario. KM0 works at the opposite scale: PTAs, neighbours, SMEs, and services you can touch. The same question applies in small.

  • Data nearby: mail, cloud, and chat at mail.km0digital.com and cloud.km0digital.com, not in one advertising profile.
  • Local agents: assistants that run next to your files, not sending every minute to the model of the week.
  • Transparency: knowing where the rules live (contracts, lists, minutes) before delegating decisions to a black box.
  • Community: growing through referrals and associations, as in meet 6, instead of feeding someone else's attention algorithms.
Open question

If tomorrow's bureaucracy speaks fluently with AI, do you want that conversation on servers your PTA controls, or on a platform that monetises every form?

Verification

How to follow up

  1. Video: watch the YouTube video and note which part feels closest to your work or association.
  2. Series: read day 13 (meet 6) for context on local AI and associations.
  3. Product: try webmail or KM0 Cloud if you want to contrast «distant AI» with «your own infrastructure».
  4. Ideas: tell us via the ideas form or contact which AI topic you would like us to cover in a meet.

Series

Previous days

Day 13 covered meet 6 on visibility and associations; day 12 the client talk on UX; day 11 documented KM0 Mail. Day 15 explains the expired register-api Graph token and automated renewal. Watch the video, share your reading, and follow the series.