The AI Learning Hub Journal

Jobs Now and Jobs Emerging

Jobs around AI, grouped by what you would actually dono salaries and no forecasts — just the shape of where the work sitsBUILDING ITthe smallest group, and the loudestEngineeringtraining, serving and toolingDatacollection, labelling, qualityResearchnew methods, and testing themInfrastructurethe machines and networks under itAPPLYING IT IN A FIELDyour subject, plus the toolHealthimaging, triage, notes, recordsLawdiscovery, drafting, researchDesign and mediaconcept, production, editingEducationmaterials, tutoring, assessmentGOVERNING ITdeciding what is allowedSafety and testingred-teaming, evals, guardrailsPolicyrules, standards, public interestAudit and assurancechecking claims against realityLaw and ethicsliability, rights, consentAND THE GROUP ALMOST EVERYONE ACTUALLY LANDS INOrdinary jobs, where the tool is part of the working day rather than the job titleNursingnotes and handoverTeachingplanning and markingTradesquotes and schedulingRetail and adminrotas, stock, emailJournalismresearch and checkingMost people will never train a model or write the rules that govern one.They will use these tools inside a job that already existed before them.Pick the field you care about first — the tool is something you add on top of it
Most careers around AI are not AI jobs — they are ordinary jobs where the tool became part of the day

Roles That Already Exist

Beyond research scientists, which is a small and highly specialised field, there is a large practical layer. Machine learning and AI engineers build and deploy systems. Data engineers make the data usable, which is most of the actual work in most organisations. AI product managers decide what gets built and for whom. Evaluation specialists design ways to measure whether a system is actually working, a role that barely existed a few years ago and is now central. Policy, governance and compliance specialists translate between technical reality and law. Most of these are not research jobs and do not require a doctorate.

  • ML and AI engineering: building, deploying and maintaining systems in production
  • Data engineering: unglamorous, in constant demand, and the bottleneck almost everywhere
  • Evaluation and testing: designing measurements of whether a system actually works
  • Policy, governance and compliance: increasingly required as regulation arrives

Roles Emerging at the Boundary

The fastest growth is often at the join between AI capability and a specific domain. People who deploy systems inside real organisations and adapt them to messy reality. People who work out how a hospital, a court, a school or a factory should actually use these tools, which requires understanding the institution as much as the technology. People who audit systems for safety and bias. People who design the human-AI interaction, which is a distinct and underrated craft. The common feature is that none of them are purely technical — they need someone fluent in the technology and fluent in something else.

  • Deployment and adaptation roles: making a general system work in one messy real setting
  • Domain specialists who understand AI, in medicine, law, education, manufacturing
  • Auditing and assurance: independently testing systems for safety, bias and reliability
  • Interaction design: deciding how people and systems actually work together

The Bigger Category: Everyone Else

Do not narrow your thinking to jobs with AI in the title. The larger effect is on ordinary roles that now include AI as a tool. A journalist who uses it well beats one who does not. Same for a lawyer, a doctor, a designer, a teacher, an electrician quoting jobs, a nurse writing notes. Being the person in a normal profession who understands what these tools do and do not do is a durable advantage, and it does not require you to become an engineer. This is also the honest answer to "will AI take my job": more often it changes what the job consists of, and the people who adapt do well.

  • Most AI-related opportunity is inside existing professions, not in AI job titles
  • Being the capable AI user on a normal team is a real and immediate advantage
  • You do not need to build models to benefit substantially from understanding them
  • The trades and hands-on work are considerably less exposed than office work

Prefer slides, quizzes, and saved progress? Read this lesson in the library — free, no sign-up.