When the machines do the work, who learns the trade?

As technology takes on more processing, analysis and execution in insurance, Sharma sees the skill mix broadening. Domain expertise now has to sit alongside technology fluency, data literacy, customer understanding and sound judgment, and roles increasingly cut across technology, operations, product and the business.

For engineers, that means producing more code stops being the differentiator. What matters is deciding what should be built, understanding how systems connect, spotting risk and judging whether AI output is actually fit for purpose. In insurance, she adds, that includes understanding the policyholder, the product, the regulatory environment and the consequences when a system gets something wrong. AI doesn't lower the bar for technical talent; it raises it.

Sharma names the paradox directly: the routine work organizations are automating is often the same work through which juniors built their foundations. Engineers learned by debugging; analysts learned by working through data. If AI removes those early repetitions, learning can't be left to experience alone. She points to simulations, shadowing, rotations, structured problem-solving and reviewing AI-generated work, along with chances to do some things the harder way before relying on the shortcut.

The same thinking extends to how HR reads roles. Automating tasks within a job doesn't necessarily make the role smaller; what remains may be more complex, built around interpreting outputs, handling exceptions and owning outcomes. Sharma argues that roles should be defined by the problems people solve and the decisions they own, not just the tasks they perform. She also makes the case for strong individual contributor paths, where growth is measured by the scale and complexity of the problems someone can solve rather than how many people report to them.