Local controlled artifact · owns this syllabus and release record · no external authority claimed.
Identity · context · boundary
One independent course, two declared contexts
This page separates the course object from the public research and practice contexts that informed its questions. Context establishes why the topic matters; it does not establish an approved offering, affiliation, endorsement, or empirical result.
- Course object
- Independent working curriculum
- Context pages
- Checked 25 August 2026 · identity-only
- Offering status
- Not announced
Source ledger · checked 25 August 2026
Know which object is speaking before using its claim
cuhkgeo.com ↗ · identity and research-theme framing only.
frontmind.net ↗ · public practice narrative only.
School research page ↗ · broader academic context. Tongxin Li faculty profile ↗ · verifies faculty role and listed research fields only; neither page verifies the named AI Decision Lab, this course, or a partnership.
Object map
Three objects remain separate by design
A professional research-to-practice curriculum must show where a statement comes from before it decides how that statement may be used.
GEO Advanced Studies
Owns the syllabus, learning outcomes, labs, assessment logic, and release record.
- Status
- Independent preview
- Authority
- Local controlled artifacts
cuhkgeo.com
The subject-published site describes a CUHK–Shenzhen AI Decision Lab and frames GEO alongside agentic decisions, trustworthy AI, control, and optimization.
Inspect the subject-published page ↗FrontMind
Its public site presents an enterprise AI path organized around understanding, growth, and embedding, with GEO positioned inside a wider delivery system.
Inspect the company-published page ↗Academic context
Four public themes become four research questions
CUHK–Shenzhen’s official School of Data Science pages establish the broader academic background. The four themes below come separately from cuhkgeo.com’s own public description and are used only as a question-generating frame; the course supplies its own evidence rules, examples, visuals, labs, and assessment decisions.
- 01Observe · Weeks 01–04
Generative Engine Optimization
How do discovery, retrieval, generation, citation, and recommendation create separate visibility outcomes?
- 02Explain · Weeks 05–08
Agents and online decisions
How should an intervention respond to feedback, drift, competition, and delayed consequences?
- 03Intervene · Weeks 09–12
Trustworthy and learning-enhanced AI
Which claims survive calibration, adversarial pressure, missing evidence, and human review?
- 04Validate · Weeks 13–16
Control, optimization, and theory
Which objectives, constraints, stopping rules, and rollback paths make an action governable?
Practice context
A company narrative becomes an auditable translation path
The course does not teach a product funnel as a scientific law. It turns each stage into a set of falsifiable questions and release conditions.
- 01
Understand
The public FrontMind narrative begins with how AI systems understand an enterprise.
COURSE TRANSLATIONThe course converts that proposition into entity, source, claim, retrieval, and attribution audits.
- 02
Grow
The public narrative then connects AI-mediated discovery to growth processes.
COURSE TRANSLATIONThe course requires an estimand, comparator, repeated measurement, adverse outcomes, and a causal ceiling before any growth claim.
- 03
Embed
The public narrative ends with enterprise workflows and AI deployment.
COURSE TRANSLATIONThe course treats integration as a governed release problem: permissions, logs, human takeover, rollback, and incident response.
Evidence firewall
Context can shape a question without raising its evidence ceiling
Permitted use
- Explain why research, decision science, and enterprise practice meet in GEO.
- Generate curriculum questions, cases, controls, and comparison axes.
- Link readers to attributed public descriptions for publisher-aware identity inspection.
Not established
- An approved CUHK–Shenzhen course, appointment, provider, or co-brand.
- Independent validation of laboratory counts, company claims, or product effects.
- A stable mechanism for a closed platform or permission to generalize marketing claims.
Verification caution
Public descriptions are dated inputs, not inherited authority
Continue the course