AI can write and code, but can it dive? 5 jobs AI may struggle to replace - India Today

TEXT START: From underwater repairs to rescuing wildlife and creating fragrances, these five unusual careers demand skills AI may struggle to replicate.

The Dissection

This is a containment narrative disguised as career advice. It identifies five physical or sensory tasks that are difficult to automate today, then quietly treats task-level resistance as occupation-level security.

The article’s real function is to preserve the old career-planning fantasy: find the right niche and remain economically safe. It converts a systemic labor-market rupture into an individual skill-selection problem. The five jobs are not evidence that the employment system survives. They are examples of the thin residue left after automation removes the scalable, repeatable, cognitive, and coordination-heavy portions of work.

The salary discussion exposes the weakness. Several roles lack reliable entry-level figures because they are fragmented, temporary, project-based, or poorly standardized. The Asiatic Society example pays ₹24,356 per month; the fragrance figure is an individual report of roughly ₹3 lakh annually. These are not sovereign escape routes. They are precarious servitor niches with weak bargaining power.

The Core Fallacy

The central error is confusing “AI cannot perform every physical action” with “AI cannot destroy the economic value of the occupation.” It does not need to replace every diver, conservator, rescuer, perfumer, or rope technician. It only needs to reduce the number required, automate the surrounding cognitive work, and weaken the remaining workers’ pricing power.

A commercial diver can remain physically necessary while AI handles inspection, planning, documentation, risk analysis, scheduling, and remote supervision. A conservator may still touch the artifact, but AI can narrow diagnosis, recommend treatments, standardize workflows, and concentrate authority in fewer experts. A perfumer’s human nose may survive as a branding and approval layer while formulation becomes algorithmically generated and tested at scale.

Physical unpredictability is a lag defense, not a permanent moat. Robotics, teleoperation, specialized tooling, autonomous inspection, better materials, and redesigned infrastructure attack the hardest tasks gradually. Even where full replacement remains uneconomic, partial automation produces the same structural result: fewer workers serving more output.

The article also commits a scale error. Five unusual occupations can remain viable for a small minority while the majority lose access to economically necessary labor. Niche persistence does not refute Productive Participation Collapse.

Hidden Assumptions

• That human intervention remains necessary at the same volume after AI and robotics absorb surrounding tasks.
• That “hard to automate completely” means economically secure.
• That unpredictable environments cannot be instrumented, modeled, or redesigned.
• That certification and safety rules will protect wages rather than merely restrict entry while firms reduce headcount.
• That human judgment cannot be concentrated into a smaller supervisory class.
• That sensory creativity, animal handling, or manual dexterity automatically commands strong compensation.
• That a job’s continued existence benefits the typical worker rather than a small number of highly trusted specialists.
• That the labor market will preserve mass participation merely because some tasks remain too expensive to automate.

These assumptions smuggle continuity into a system whose central wage-to-consumption circuit is being severed. The article never addresses ownership, bargaining power, demand volume, or who captures the productivity gains.

Social Function

Primary classification: ideological anesthetic and transition management, with a partial truth.

The partial truth is real: hands-on work in dangerous, irregular environments is harder to automate than routine office work. The anesthetic is treating that delay as safety. The article encourages readers to hunt for “AI-proof” jobs instead of recognizing that the durable outcome may be a smaller number of human specialists surrounded by automated systems and a much larger population excluded from productive participation.

It also performs elite self-exoneration. If displaced workers are told to retrain as divers, conservators, wildlife rescuers, perfumers, or rope technicians, structural failure can be reframed as insufficient imagination or adaptability. The system’s inability to provide mass employment disappears behind a list of exotic exceptions.

The Verdict

The article is not wrong that these jobs are difficult to automate immediately. It is wrong about what that fact means. They are not escape hatches from the Discontinuity; they are lagging niches inside it.

Their survival depends on physical presence, regulation, trust, local knowledge, and messy environments—the exact forms of friction that delay automation. None restores mass productive participation. At best, they provide limited servitor positions for workers who remain indispensable to owners of automated capital. The article mistakes the persistence of difficult tasks for the survival of the post-WWII economic order.