Nine Cards

For the AI Co-Scientist

Initiative × Scope — Find the kind of AI agent that fits how you work.

Take a five-minute survey for a chance to receive a $200 USD-equivalent electronic incentive. AI is already changing how you work. Take the survey to find out how you collaborate with AI, and which AI partnership style best supports your scientific work.

Take the Test

The grid below has three rows (the scope of work the AI handles) and three columns (the initiative the AI takes). Each cell is one type of AI agent for data-science research, mapped to a Tarot archetype and a scientific instrument from the natural-philosophy tradition. Click any card to read its description.

The routing logic is described under each card's How you got here section.

User—Initiated

Mixed—Initiated

System—Initiated

Narrow

I · The Magician of the Compass — Lapidary

II · Temperance of the Sundial — Adjuster

III · Justice of the Plumb-Line — Inspector

Broad

IV · The Empress of the Alembic — Drafter

V · The Lovers of the Astrolabe — Collaborator

VI · The Wheel of the Orrery — Conductor

Whole Task

VIII · The High Priestess of the Microscope — Co-Investigator

IX · The Hermit of the Telescope — Successor

The Two Axes, Defined

Axis

Definition

Survey Signal

Scope

How much of a task the AI handles in one invocation. Narrow (a single function call) → Broad (a coherent multi-step subtask) → Whole (end-to-end goal pursuit).

Q3 (current uses) and Q10 (what would justify setup), averaged. Q1c competence (more competent → narrower) and Q2 frequency (daily → broader) modulate.

Initiative

Who proposes the next move. User-initiated (system only responds when invoked) → Mixed (either party can propose) → System-initiated (system drives, user reacts).

Q11 (top desired qualities), Q4 (current frustrations), Q10 contributions, summed. Q1c and Q2 modulate.

Honest Notes on the Typology

Cells I (Magician) through VI (Wheel) describe AI tools that exist in current daily research practice.
Cell VIII (Priestess) is the survey's central aspirational target — the "AI co-scientist" as currently positioned by Sakana, FutureHouse, and Google AI Co-Scientist.
Cells VII (Fool) and IX (Hermit) describe whole-task agency: VII as currently shipping autonomous-agent products (Devin, Manus); IX as mostly aspirational, requiring trust infrastructure that does not yet exist.
The typology is descriptive, not prescriptive — the goal is to make different positions toward AI tools legible and discussable, not to rank them.