Mag.AI-Science
Mag.AI-Science¶
Castalia Institute — Magister of Artificial Intelligence in Science
Science is not only what you claim—it is what you can reproduce, measure, and defend under scrutiny.
If you’re new here¶
Mag.AI-S is 18 courses over three terms: you move from measurement, reproducibility, and evidence systems through experiment design, simulation, and open science, to strategy, governance, and a capstone. You move from intuition toward executable hypotheses—workflows you can stress-test, toolchains you can scope, and artifacts you can defend to faculty and reviewers. You are not graded on slide fluency—you build artifacts (notebooks, simulations, documented lineage) that others can inspect.
Where work happens: readings and maps live in these course books; hands-on work runs in iNQspace (Castalia’s lab—notebooks, simulations, and tracked lineage). Executable articles (Jupyter Book per course) are the public map; SAMWISE and faculty defense sit where your course assigns them.
Words you’ll see¶
| Term | Meaning here |
|---|---|
| Magisterium | Castalia’s credential system—earned through demonstrated work, not attendance alone. |
| MCP | Model Context Protocol—connect AI workflows to instruments, literature, and lab systems with explicit scope; policy is in MCP & scientific tools. |
| Artifact | Something you built that someone else can review—code, notebook, sim, instrument trace. |
| SAMWISE | Structured reflection and critique in the stack—not a generic “journal.” |
| AINS-S#### | Course codes (like AINS-S6001); use them to find your module in the catalog. |
What makes this different¶
| Typical AI training | Mag.AI-S |
|---|---|
| Prompt tricks and one-off demos | Testable world models, measurement, and uncertainty |
| Disconnected from your lab or data stack | MCP-aware workflows tied to instruments and systems (where the program requires it) |
| “Understand AI” as vocabulary | AI literacy spine: evaluation, grounding, retrieval, oversight—embedded in scientific artifacts |
| Slides without runnable lineage | Executable articles: Jupyter Book sites per course; notebooks in iNQspace (policy) |
| Certificate as completion | Magisterium credential: artifacts, defenses, and a thesis-grade deployment |
What you’ll leave with¶
Models you can run — hypotheses, uncertainty, and evidence chains as mechanics, not metaphors
Reproducibility habits — provenance, documentation, and honest limitation statements
Tool fluency — connecting AI workflows to scientific systems with explicit scopes (MCP & scientific tools)
A graduate toolbox — documented lineage from notebooks to validated systems (certificate & graduate toolbox)
Who this is for¶
Scientists and engineers who own inference, not just implementation
Lab and platform leads who need alignment between instruments, data, and claims
Builders who want AI as an instrument—auditable, bounded, and improvable
This is not a light tips course: expect rigor, documentation, and critique. It is built so you always know what to read next and where to build—start below, not by guessing.
Your first steps¶
Program overview · Curriculum at a glance
How the program thinks and how the three terms fit together.
AINS-S6001 — first course book
Read the syllabus, then lectures → notebooks in the sidebar (recommended entry for Term I).
Assessment · MCP & scientific tools
When a doc points you to policy or grading, start here.
Full catalog with links to every module.
Navigate this hub¶
This page is the program hub (big-picture story, policies, links). Each course has its own book under /ains-s####/. Use the left sidebar or jump below.
Orientation
Program overview · Curriculum at a glance · Certificate & credential
Deep reference
Policies
- Mag.AI-Science
- Course sites
- Orientation
- Program reference
Legal¶
Note
The Castalia Institute Magisterium confers proprietary credentials based on demonstrated work and evaluation. These credentials are not accredited academic degrees and do not confer professional licensure.