AI-Era Skills
Which UCM courses actually prepare Manx students for an AI-native career? We classify 189 active programmes into four honest buckets — from "explicitly teaches AI tooling" through "distinctly human, AI-resistant" to "at-risk routine work" — so students, Treasury and UCM can make informed calls.
Explicitly AI-Native (11)
Course name or field directly teaches AI tools, data science, cyber security, programming, cloud, automation, prompting. These graduates step straight into AI-era roles.
AI-Augmented (35)
Course leads to occupations the Anthropic Economic Index rates as augmentation-dominant — AI makes the graduate more productive, humans stay in the loop.
Distinctly Human (57)
Course leads to occupations whose core tasks are hard for AI to touch — personal service, physical trades, care, supervised practice. AI-resistant regardless of the era.
AI-Vulnerable (86)
Course leads to occupations with high automation exposure and limited augmentation upside. Students should know this before enrolling — not a reason to cancel, but a reason to pair with AI-fluency add-ons.
How the classification works
Classification priority — the first match wins:
- Explicitly AI-Native — course name or primary field matches a pattern like artificial intelligence, machine learning, data science, cyber security, cloud, DevOps, coding, programming, software, analytics, automation, robotics, generative, LLM, prompt engineering.
- AI-Augmented — matched SOCs have Anthropic Economic Index augmentation share ≥ 35% AND augmentation > automation. AI assists, human stays central.
- Distinctly Human — matched SOCs retain ≥ 30% distinctly-human task share, OR the course is a trade apprenticeship (AP category) with sub-35% automation. AI has limited operational reach.
- AI-Vulnerable — matched SOCs have automation share ≥ 40% AND augmentation < 30%. Routine tasks are automatable and AI doesn't help much with what's left.
Courses with no Anthropic EI data for their matched SOCs are excluded to keep the signal clean. The page is a planning aid, not a prescription — a "distinctly human" course is not automatically "safer" than an "AI-vulnerable" course if the market for it is shrinking.
