Back to the reading room

/ Information & trust

Nexus

A Brief History of Information Networks from the Stone Age to AI

4 / 5CEO rating

Recommended by Abdollah Dehnashi

A timely examination of information, power, and the responsibilities that come with increasingly capable AI.

The review

In Nexus, Harari examines information networks across history and asks how they shape cooperation, institutions, and power. His central concern is not simply the amount of information available, but the relationship between information and truth. AI gives these questions new urgency because it can participate in producing and shaping information at scale.

The enterprise implication is practical: more data and faster answers do not automatically produce better judgment. A leadership team needs to know where an answer came from, which assumptions it carries, and how an error can be corrected. That makes the book relevant to organizations building assistants, knowledge systems, or automated decision workflows.

Consider an illustrative internal assistant that gives a confident answer from an outdated policy document. The immediate error is a wrong answer; the deeper issue is a network that has no reliable way to withdraw the old source, identify affected users, and circulate a correction. A useful reading of Nexus therefore moves beyond checking one response toward examining how information acquires authority. In everyday life, forwarding a persuasive message raises a smaller version of the same question: what makes it worth trusting, and what happens if it is wrong?

Its value lies in the breadth of the questions it raises. The historical and political scope can be provocative, and readers should examine its arguments rather than treat them as operating instructions. For a business audience, the strongest use is to turn the debate into specific questions about information ownership, review, correction, and accountability.

The following are our practical applications of the book’s ideas. Choose one action to try this week, then reflect on what changed.

IDEAS INTO ACTION · AT WORK

Three things to take to work

  1. Trace one answer back to evidence

    Pick a decision supported by an AI response and follow its sources to their owners and dates. Record any gap between what the evidence says and what the response claims.

  2. Practise a correction end to end

    Use an outdated document as an illustrative test case. Identify who replaces it, which systems consume it, and how people who relied on the old version would learn about the correction.

  3. Give challenges a visible route

    Make it clear where a colleague can question an automated answer and who must respond. Review unresolved challenges so repeated concerns become improvements to the information process.

BEYOND THE WORKING DAY

Three things to take into life

  1. Pause before passing a claim on

    Before forwarding a surprising message, open the original source and check its date and context. A strong emotional reaction is a reason to investigate, not evidence that the claim is true.

  2. Make a correction easy to find

    If you share something inaccurate, correct it in the same conversation or channel. Say what changed so others can update their understanding without having to reconstruct the mistake.

  3. Choose the quality of your information

    Review one habitual source of news or advice. Ask whether it shows evidence, acknowledges uncertainty, and corrects errors; use those behaviours to judge its usefulness.

A question for your team

If an AI-generated answer is wrong, can your organization identify its source and correct its downstream effects?

FROM READING TO PRACTICE

How Genellipse can help

Genellipse can help map the information journey behind a business workflow. GeneData supports the discussion about data foundations, Cortex AI connects that information to governed agent activity, and GeneComply can be considered for evidence and control workflows. We work with partners to define ownership, review, and escalation around their specific requirements.

Explore the related service

These reviews were prepared retrospectively in September 2026. Month labels organize the reading series and are not original publication dates. Personal ratings are provided by Abdollah Dehnashi; the reviews and practical applications are prepared for CEO Corner.