Artificial Intelligence reading list

**The full list, in order:** (we’ll come back to this / just mapping out our thoughts)

—

**1. Sapiens** (Harari, 2011)

What made humans the dominant species. The answer — shared fictions, abstraction, cooperation at scale — sets up why the AI question matters. If our power comes from our ability to create and coordinate through stories and symbols, what happens when something else can do that?

—

**2. Artificial Intelligence: The Very Idea** (Haugeland, 1989)

What the AI project was actually trying to do. Haugeland explains GOFAI (Good Old-Fashioned AI) charitably and clearly: the bet that intelligence is symbol manipulation, that mind is computation. You need this to understand why neural networks felt like a rupture — and why the old debates don’t quite map onto what’s happening now.

—

**3. Mind Over Machine** (Dreyfus, 1986)

Why the classical project couldn’t work. The phenomenological critique: human expertise isn’t rules, it’s embodied intuition developed through experience. Dreyfus was “wrong” about chess and expert systems on the surface, but his deeper argument about cognition has never been answered.

—

**4. Atlas of AI** (Crawford, 2021)

The material reality underneath. Lithium mines, water consumption, data labelers in Kenya. Pulls the discourse out of abstraction and into the physical world. The critical lens that asks: who pays for this, who benefits, who decides?

—

**5. Algorithms of Oppression** (Noble, 2018)

*Optional — skim if Crawford covered enough ground.*

How “neutral” systems encode existing power structures. Focused on search specifically. If you already have a solid inequality/feminism framework, the specific examples may be more useful than the theory.

—

**6. The Alignment Problem** (Christian, 2020)

Why it’s hard to make AI do what we actually want. The safety question, treated rigorously without polemic. Bridges the critical tradition and the builders’ concerns. This is the book that helps you understand why serious people are worried without telling you the sky is falling.

—

**7. Empire of AI** (Hao, 2025)

The current moment. Investigative journalism on OpenAI, Altman, the board coup, the safety team departures. Deeply reported, 260 interviews. Tells you how we got *here* — the specific corporate and personal dynamics driving the race.

—

**8. If Anyone Builds It, Everyone Dies** (Yudkowsky/Soares, 2025)

The maximalist alarm. What if superintelligence actually arrives? The strongest form of the doom argument. Read it after Christian so you can evaluate whether the certainty is warranted.

—

**9. Homo Deus** (Harari, 2017)

The capstone. Step back: where does all this leave humanity? Harari speculates about a future where algorithms know us better than we know ourselves, where “humanism” as an organizing principle may be ending. After reading the technical, critical, and journalistic accounts, this is the long view out.

—

**10. AI Engineering** (Huyen)

*Defer until you want to build.*

Practical, technical, about making ML systems work in production. Will age fastest. Use as a reference when you’re actually doing the work.

—

If you only read one: Sapiens

If only three:

Sapiens — what made us this
The Alignment Problem — the current problem
Homo Deus — what we might become

If only five:

Sapiens
Haugeland
Dreyfus
The Alignment Problem
Homo Deus