Augmentation Before Automation: What AI Really Does to Jobs

The public debate about AI and employment is mostly conducted at the level of entire occupations: will AI replace accountants, will it replace lawyers, will it replace radiologists. That framing makes for a dramatic headline, but it rarely matches what actually happens inside a real organisation in the first few years of adoption. What we see, almost without exception, is change at the level of tasks within a job, not the wholesale disappearance of the job itself.

Take a finance team preparing monthly management accounts. The job has never really been “produce the accounts”; it’s a bundle of tasks that includes reconciling exports from two or three systems that were never designed to talk to each other, chasing down which of two conflicting numbers is correct, formatting a workbook so the board can read it, and then, right at the end, actually interpreting what the numbers mean for the business. AI-assisted automation is very good at the first three of those tasks and not particularly good at the fourth. The practical effect, in the engagements we’ve run, is not that the finance role disappears; it’s that the ratio of time spent shifts hard away from reconciliation and formatting and toward interpretation and judgement. That is a better job, not a smaller one, but it does mean the skills that make someone valuable in that seat are changing under their feet.

Augmentation is a management choice, not a technology default

It’s worth being direct about something the more utopian AI commentary tends to skip: augmentation over automation is not guaranteed by the technology. It’s a choice made by whoever is designing the workflow. The same automation that frees an analyst to spend more time on judgement can just as easily be used to justify cutting the headcount and keeping the workload the same. Both are real options with the same underlying tools. Which one an organisation ends up with depends on decisions made well before any AI system is switched on: what the business is actually trying to get more of, capacity, quality, speed, headcount reduction, and how honestly that goal is stated to the people whose jobs are affected.

For businesses in Zimbabwe and the wider region, where skilled staff are often a genuine competitive advantage and not easily replaced, the augmentation path tends to be the economically sound one as well as the more humane one. Losing an experienced accounts clerk or a trained radiographer because their routine tasks got automated, only to discover eighteen months later that the judgement work they used to do quietly alongside the routine work is now missing too, is a expensive way to find out that a job was never as separable as an org chart suggested.

The practical takeaway for leadership is to ask, before any AI project starts, what a given role’s staff will spend their newly freed time doing, and to make that answer explicit rather than assuming it will sort itself out. If there’s no good answer to that question, the project probably isn’t ready to start.