The Argument Tree

What this is

The first map asks why AI development should keep its speed and concludes that the open question is its direction. The second asks why that direction needs a governor, a check built into the systems themselves, and why it has to be built now. The third sets out what any such governor would have to meet. Each map's result is a premise the next one uses.

Every step has a canonical form and every sentence a premise relies on is stated. Each part of a map sits on its own coloured panel, labelled with what it returns; the heavily bordered boxes at the bottom are the spine that combines the parts.

How to read it

• Pick a map across the top. The first two open on an Overview: a coloured box for the result of each part, with its steps folded away, and the spine below. Click a box to unfold its part; Whole tree unfolds everything.
• The third map opens on a table of its six small maps: what each establishes, its steps and what is still open. Open a row to see that tree.
• Choose a part in the key or the Show menu to see it alone; click any box to read its step in full.
• In a step, underlined terms open their definition; evidence tags open what supports a premise; ? objection tags open the reply as its own small map.
• Show formal version puts each step in its logical form, with the rule it uses.

How it is checked

• 70 steps across 204 statements, every one passing the page's word-for-word check: 8 in the first map, 49 in the second map and its replies, 13 in the third.
• Each step names its rule of inference; a premise that is merely implied by the text is shown with a dashed border and marked as implicit.
• Each premise carries a status: shown (the book gives evidence), probable, hypothesis, assumption (granted by the book's opponents), or definition.
• The second map and five of the six criteria maps are also written in Lean 4, every premise an axiom and every step a theorem; the file compiles with no errors and is in the File tab with the data for the whole tree.