(A More) Objective View of AI

^{_{AI Can Help or Harm the Planet. It's Up to Us}} (^{_{original image from Nasa}})

The profound influence of AI on technology, human culture, and society is undeniable. Opinions about AI range from predictions of imminent societal or economic collapse to, at the other extreme, excessive excitement and overreliance.

These open-source notes aim to offer a more balanced and objective view of Artificial Intelligence (AI).

Every claim is backed by references from peer-reviewed research, industry reports, and recognized practitioners.

Audience:

• People who want to understand AI's real capabilities and limitations through scientific evidence (like me).
• People curious about what AI actually is (and isn't) from a cognitive and computational perspective.
• Software engineers (often overexcited) who want to use AI as a tool, not a replacement.

If you enjoy these notes or you find them useful, please add a Star. (This will help others discover them)

\* These notes are a living document and will be updated with new research findings and community contributions.

Table of Contents

0. Preface and Introduction (pdf)

1. Biological and Artificial Intelligence (pdf)

A broad examination of AI from scientific, cognitive, and social perspectives.

• Introduction
• AI and Human Brain are Structurally Different
• AI and Human Brain are Functionally Different
• The Illusion of Consciousness

- Who is being deceived?

2. AI Limitations (pdf)

Coverage of AI limitations from a technical and social perspective.

• AI Technical Limitations

- Hallucinations are Mathematically Inevitable - Prompt Sensitivity - Generalization - Creativity - Production Quality - Causality and Real-World Understanding - Non-Determinism - Introspection - Quality Fluctuation

• AI "Social" Limitations

- Intellectual Property - Sycophancy - Alignment - Social Coherence - Security Risks - Liability

3. Implications for the Future, Part 1: Growth, Productivity, and Employment (pdf)

AI consequences on growth, productivity, and employment.

• Introduction
• Unlikely Exponential Growth
• Productivity Gain Uncertainty
• Employment

- Employment Data - Employment and Early Career Jobs - Employment Predictions - Employment and Productivity - Economic Impact

4. Implications for the Future, Part 2: Positive and Negative Impacts (pdf)

AI consequences on society, science, health, politics, and the environment.

• Positive Impacts

- Protein Structure Prediction - Material Discovery - Weather Forecasting - Healthcare - Ancient Languages - Learning - Soften Polarization

• Negative Impacts

- Environmental Risks - Cognitive Offloading and Deskilling - Echo Chamber - Health Risks - Inequality - User Manipulation - Social Risks - Religious - Political Influence

5. Software Development in the Age of AI, Part 1: Implications of AI-Generated Code (pdf)

A practical look at the effects of AI-generated code on software work.

• Software Developer Employment
• Productivity
• Deskilling
• AI Addiction
• Technical Debt
• Security Vulnerabilities
• Maintainability
• Code Review
• Comprehension Debt
• Cost

6. Software Development in the Age of AI, Part 2: Limitations and Engineering Practices (pdf)

A practical look at AI limitations and engineering practices.

• Limitations of AI-Generated Code

- Slot Machine and the Illusion of Control - Generalization - Creativity - The Illusion of Competence - Long-Term Tasks - The "Last Mile" Problem

• AI as a Software Engineering Tool

- Code Generation is NOT Software Engineering - The Role of Human Expertise

• Engineering Practices in the Age of AI

Format

The notes are written in Typst, a modern markup language, alternative to LaTeX, for creating documents. They are compiled to PDF and HTML.

Contributing

Please see the CONTRIBUTING.md file for more information.

Further References

• All about AI by Andrew Perfors.
• Modern-Day Oracle or Bullshit Machines by Carl T. Bergstrom and Jevin D. West.
• The Dangerous Illusion of AI Coding? by Jeremy Howard.

License

This repository is dual-licensed:

• Course Content: All written materials, slides, and images are licensed under Creative Commons Attribution 4.0 International (CC BY-SA 4.0).
• Source Code: All code examples and scripts are licensed under the MIT License.

Author

Federico Busato, federico-busato.github.io

• 🌐 LinkedIn: www.linkedin.com/in/federico-busato/
• 🦋 Bluesky: fbusato.bsky.social