New AI Roles
New AI roles name the emerging category of jobs organized around AI systems: prompt engineers who craft inputs, AI trainers who shape outputs, human-in-the-loop specialists who correct errors, alignment researchers who address safety, AI ethicists who articulate principles. These roles are frequently presented as evidence that AI creates more work than it eliminates and as examples of the new opportunities the transition affords. Read through Boltanski's framework, they reveal a more complicated pattern: they are the projective city's latest instantiation, incorporating the vocabulary of technical sophistication while reproducing the structural precarity that has characterized platform-mediated work for a generation.
In the AI Story
Each role exhibits a characteristic pattern. The title sounds technical and specialized, suggesting durable expertise. The requirements are in fact loose enough that workers can be replaced rapidly by others who have picked up comparable skills. The compensation reflects the glamour of AI more than the durability of the skill: early entrants are paid well, but the premium erodes as the skill diffuses. The employment arrangement is often contractual, project-based, or outsourced, without the benefits, stability, or representation that direct employment would provide.
The prompt engineer is the paradigm case. The role emerged in 2022-2023 as something genuinely novel — the capacity to elicit quality output from language models required specific expertise that was in short supply. By 2025, the skill had diffused broadly; tools automated much of what prompt engineers did; and the role was increasingly absorbed into other job descriptions. Workers who had built careers around prompt engineering found themselves explaining, three years in, that their specialization no longer existed as a distinct role.
The pattern is not accidental. It reflects the projective city's characteristic rhythm: rapid creation of new roles that serve current needs, rapid obsolescence as those needs evolve, rapid reinvention demanded of workers who must continuously update their project portfolios. The mobility the order celebrates is the structural reality of not being able to stay put.
The critique is not that these roles are unreal or that the people doing them are not contributing. They are, and they are. The critique is that presenting these roles as the solution to AI-driven displacement mistakes the surface for the structure. Behind each new role is the same question: who captures the surplus from AI-driven productivity gains? If the answer is capital, the new roles are sophisticated-sounding precarity. If the answer includes labor, the new roles are a step forward. The discourse surrounding the roles systematically obscures this question.
Origin
The roles emerged in the 2022-2025 period as commercial AI capability expanded and firms discovered they needed humans to shape AI systems. The category was immediately absorbed into management literature as evidence of AI's job-creating potential, reproducing the rhetorical pattern Boltanski documented across previous technological transitions.
Key Ideas
Projective-city reproduction. New AI roles instantiate the order of worth that celebrates mobility, connection, and adaptability.
Rapid obsolescence. The roles emerge, diffuse, and dissolve on timescales of 2-3 years, preventing accumulation of durable expertise.
Contractual precarity. Most such roles are project-based, contract, or outsourced rather than stable employment.
Rhetorical function. The existence of new roles is used to deflect critique of AI-driven displacement, regardless of the roles' actual economic significance.
Surplus question hidden. The distribution of gains from AI-driven productivity is obscured by focus on job counts.
Appears in the Orange Pill Cycle
Further reading
- Mary Gray and Siddharth Suri, Ghost Work (Houghton Mifflin, 2019)
- Kate Crawford, Atlas of AI (Yale, 2021)
- James Muldoon et al., 'Picking up the Slack' (Ai & Society, 2023)