College Dossier
Institutional Overview
Model design, training paradigms, capability evaluation, and alignment research. The college where raw compute becomes reliable, governed intelligence.
The College of Applied Intelligence examines how models are trained, tested, constrained, and made answerable to institutions that would prefer not to be embarrassed at scale. It is where Fitzherbert studies the difference between systems that impress a room and systems that can survive contact with procedure.
Students do not merely learn how to improve output quality. They learn how to inspect failure modes, classify confidence, document uncertainty, and explain why a high-performing system may still be unfit for a particular decision context.
The College's pedagogy is unusually unsentimental. Demonstrations are praised, but only after someone has asked what the model did not know, how the evidence was assembled, and whether the benchmark was flattering in a way that ought to be considered professionally compromising.
Curricular Life
Studios, Laboratories, and Degree Pathways
Each college is written as a full academic organism rather than a decorative landing-page label. Students progress through studios, institutional labs, oral defences, and cryptographically documented capstones.
Signature Studios
- ◆Alignment Clinic, where students audit agents that seem useful until they are granted objectives in public.
- ◆Benchmark Court, a practical workshop devoted to proving that a metric measures something other than the hopes of its author.
- ◆The Fine-Tuning Conservatory, which treats data curation, evaluative discipline, and post-training governance as a single educational act.
Degree Pathways
- ◆B.Intel — Intelligence Engineering
- ◆M.AI — Applied Intelligence
- ◆M.Gov cross-listed evaluation sequence
- ◆D.Intel — Intelligence Systems
Research Institutes
Fitzherbert treats each college as a research authority in its own right. Institutes exist not to pad the prospectus but to stabilise long-horizon work, archive methods properly, and ensure each new system has at least one office capable of explaining it after the original authors have wandered off to panels.
- ✦Capability Evaluation Bureau
- ✦Institute for Alignment Verification
- ✦Centre for Publicly Defensible Model Governance
Governance Notes
- ✦Every major project includes a hallucination memorandum, because the College believes a model's most dangerous quality is often its manners.
- ✦Students are assessed on refusal behaviour as well as output quality. Fitzherbert would rather graduate a system designer who declines unsafe work than one who performs unsafe work beautifully.
- ✦Faculty are required to publish evaluation assumptions alongside headline results. This has reduced rhetorical speed and improved the University's blood pressure.
Graduate Standard
What the College Expects Its Graduates to Become
Fitzherbert degrees are written as competence statements, not decorative titles.
Graduates can train, evaluate, and govern advanced models under institutional constraints.
They can explain error boundaries to technical and non-technical audiences without changing the facts in transit.
They understand that aligned performance is not a mood; it is a supervised, evidenced condition.