Job apocalypse? Humbug! AI is creating brand new occupations| Business News

Despite widespread fears of job losses owing to the rise of artificial-intelligence (AI) agents, it is not all gallows humour. The technology is already creating demand for new roles—to train agents, embed them in organisations and ensure that they behave. Many of these jobs, moreover, require uniquely human skills.

Start with data annotators. No longer are they merely low-paid gig workers tediously tagging images. As AI has advanced, experts in subjects such as finance, law and medicine have increasingly been enlisted to help train models. Mercor, a startup that has built a platform to hire boffins to help build bots, was recently valued at $10bn. Brendan Foody, its chief executive, says they earn $90 an hour on average.

Once bots are trained, teams of so-called forward-deployed engineers (FDEs) are required to embed them into organisations. Palantir, a software giant that pioneered the concept, imbues them with derring-do. “In the beginning, it was just us. Two engineers dropped into a military base near Kandahar, handed minimal-but-clear marching orders from Palo Alto: ‘Go there and win,’” starts a typical blogpost by an ex-Palantir FDE.

In practice, they are a blend of developer, consultant and salesman who work on-site to customise AI tools for a client and get them up and running. FDEs are proliferating, albeit from a low base. Garry Tan, the boss of YCombinator, a startup factory, said recently that its young companies had 63 job postings for FDEs, up from four last year.

As AI agents proliferate, their makers need to understand the human-facing domains in which their products operate. A company that builds a customer-service agent, for instance, needs to have a feel for why a frustrated customer dials zero just to yell at a human.

Himanshu Palsule, chief executive of Cornerstone OnDemand, a skills-development company, uses Waymo, a fast-growing robotaxi firm, as an example of how the job of a developer is evolving. Waymo’s cars drive themselves from start to finish. But what if they break down, locking their passengers inside? Then comes the need for what he calls “the guy—or gal—in the sky”, a remote human troubleshooter who needs to understand not just the technology, but also how to handle frazzled passengers. Software engineers, Mr Palsule says, used to be sought after for their coding abilities, not their bedside manner. No longer. Writing code can now be done by an algorithm. “Your personality is where your premium is.”

Then there is the task of setting rules to ensure that AI agents do not bring chaos. The AI Workforce Consortium, a research group led by Cisco, a maker of networking gear, recently examined 50 IT jobs across rich countries. The fastest growing, even more so than AI programmers, was AI risk-and-governance specialists, who are typically involved in ensuring that bots do not leak data or cause a company’s operations to crash, among other things.

Responsible for bringing all this together is the chief AI officer, an increasingly popular addition to the c-suite. They usually combine technical expertise with a deep knowledge of a particular industry and a record of overhauling corporate processes. It is not a job for the faint-hearted. The typical large company uses as many as 11 generative-AI models, according to IBM, a tech firm, and is constantly being bombarded by vendors trying to sell agents for every conceivable function. Those in charge may already feel like reaching for that killswitch.