Knowingo
Try 7 days free
The cover of A Human's Guide to Machine Intelligence

A Human's Guide to Machine Intelligence

Kartik Hosanagar

Algorithms now shape what we buy, read and believe, so how do they work, why do they surprise us, and how can we stay in charge?

Get the app Read on the web

At a glance

Author
Kartik Hosanagar

Why read it

  • You will understand how a computer reaches a decision, from a simple recipe to a system that learns
  • You will see why systems built with good intentions can produce results nobody planned
  • You will know what rights and controls to ask for when an algorithm makes decisions about your life

Who it is for

  • A manager who relies on software to make decisions
  • A professional worried that AI will make choices for them
  • A curious person who uses recommendations daily and wonders why

The idea people argue about

Hosanagar's most debatable proposal is that people can retain meaningful agency while delegating more choices to algorithms. A common objection is that personalization systems learn from hidden data, optimize goals selected by firms, and shape the options users see, so individual control may be more formal than real. Hosanagar responds with a proposed bill of rights built around awareness, explanation, and control, together with stronger organizational accountability. He also shows why publishing source code alone would not solve the problem, since complex models, training data, objectives, and changing environments all affect outcomes. The proposal accepts algorithmic assistance rather than calling for its removal. Its difficulty is enforcement: the book identifies rights users should possess, but rights become effective only when institutions can inspect systems, contest harmful decisions, and impose consequences on organizations that benefit from opacity.

Three common misunderstandings

  1. An algorithm is not socially neutral merely because its operations are mathematical and its output appears numerical. Designers choose objectives, data, categories, and measures of success, while historical patterns in the data can carry unequal treatment into the output.
  2. Recommendation does not always mean deliberate manipulation by a malicious engineer following one secret commercial plan. The book shows that harmful or narrowing effects can emerge from reasonable local goals, feedback loops, proxy measures, and interactions that nobody planned in full.
  3. Transparency is not equivalent to giving every affected user a long page of unexplained technical source code alone. Hosanagar emphasizes understandable reasons, knowledge that automation is involved, practical control, and accountability appropriate to the decision and audience.

Contents

  1. Introduction
  2. Part One: The Rogue Code
  3. 1. Free Will in an Algorithmic World
  4. 2. The Law of Unanticipated Consequences
  5. Part Two: Algorithmic Thinking
  6. 3. Omelet Recipes for Computers
  7. 4. Algorithms Become Intelligent
  8. 5. Machine Learning and the Predictability-Resilience Paradox
  9. 6. The Psychology of Algorithms
  10. Part Three: Taming the Code
  11. 7. In Algorithms We Trust
  12. 8. Which Is to Be Master—algorithm or User?
  13. 9. Inside the Black Box
  14. 10. An Algorithmic Bill of Rights
  15. Conclusion: The Games Algorithms Play

More books like this