Every few weeks a post goes round claiming AI has wiped out tens of thousands of jobs in some profession. The numbers are always specific. They are almost never sourced.
So here is the opposite: what the official employment data actually records, what the research on Africa actually projects, and what that means if you are building a career in Ghana.
The short version is that the honest picture is less dramatic than the headlines and more urgent for you personally.
The job-loss numbers going round are mostly invented
Take paralegals, a favourite example because legal AI tools are genuinely impressive. Harvey, the best-known one, is used by over 142,000 legal professionals at more than 60 Am Law 100 firms, and reached $300 million in annual recurring revenue by May 2026. Real adoption, real money.
Now the employment data. The US Bureau of Labor Statistics projects paralegal employment going from 404,900 in 2025 to 403,800 in 2035. That is a net change of about 0% across a decade — roughly 1,100 fewer posts out of more than 400,000. Over the same period it expects around 40,800 openings a year, mostly replacing people who retire or move on.
What BLS does say is that demand will be limited by technology including AI — growth that would otherwise have happened does not happen. That is a real effect. It is just a completely different thing from mass layoffs, and the difference matters when you are deciding what to study.
The number that actually applies to Ghana
The research worth your attention is not about American paralegals. It is about the sector that employs people here.
In April 2025, Caribou and Genesis Analytics published Preparing for AI in the BPO and ITES Sector in Africa, in partnership with the Mastercard Foundation. Business process outsourcing and IT-enabled services employ over one million people across Africa. The headline finding:
Read that carefully: tasks, not jobs. A role made of ten tasks where four can be automated does not disappear. It changes, and usually it means fewer people are needed to do the same volume.
The detail that should concern anyone starting out is this. Entry-level jobs make up 68% of the workforce in that sector, and more than half of entry-level tasks are automatable. In finance and accounting roles, nearly two-thirds of junior-level tasks are exposed.
Entry-level work is how people get in. Data entry, first-line support, basic bookkeeping, transcription — these were the jobs you took to get experience. They are the most automatable work in the sector.
One caveat worth stating: that report covers Africa's outsourcing sector as a whole and does not single Ghana out. The pattern applies to the kind of work done here, but it is not a Ghana-specific forecast.
Where Ghana actually stands
Ghana ranked 17th globally in the 2026 Global Outsourcing Talent Index, putting it in the top 9% of countries assessed — on labour cost, available talent and English proficiency.
That is a genuinely strong position, and it cuts both ways. Being competitive at outsourcing work means Ghana is well placed to win more of it. It also means the country is exposed to exactly the automation the research describes.
There is also a sobering line in the 2026 index coverage: fewer than half of firms using AI in outsourced services have recorded meaningful productivity gains. The disruption is arriving faster than the proven benefit.
What this means if you are starting out
The useful question is not "will AI take my job". It is "which parts of the job I want are the automatable parts, and what else can I do".
| More exposed | Less exposed |
|---|---|
| Repetitive, rule-based, done the same way every time | Judgement calls where being wrong has consequences |
| Producing a first draft from a template | Checking a draft and taking responsibility for it |
| Work with no physical or local component | Work needing presence, local knowledge or trust |
| Tasks where "close enough" is acceptable | Tasks where a regulator or client demands a named human |
The pattern across every serious study is the same: the work that survives is the work where a person is accountable for the output. AI drafts; someone has to sign.
Practically, for someone in Ghana:
- Learn the verification skill, not just the tool. Being able to use an AI tool is becoming common. Being able to spot where it is confidently wrong is not.
- Get the credential that lets you sign off. Professional registration, a licence, a recognised qualification — these are what make a human legally necessary rather than merely useful.
- Do not count on entry-level work lasting as a stepping stone. If more than half of those tasks go, the ladder loses its bottom rungs. Build evidence of your work in public instead of waiting for a junior post to train you.
- Aim at what cannot be done remotely from anywhere. Work requiring local presence, local relationships or local regulatory knowledge is the hardest to automate or offshore.
How to judge the next AI-jobs headline
You will see more of these articles. Four quick checks:
- Is there a named source for the number? "127,000 positions eliminated" with no citation is fiction. Real labour statistics come from a named statistics office.
- Does it say tasks or jobs? Serious research measures task exposure. Headlines convert that into job losses, which is not the same claim.
- Is it a projection or a count? "Could be automated by 2030" is a forecast with assumptions. "Employment fell by X last year" is a measurement.
- Who benefits from the number being scary? Vendors selling AI tools and outlets selling clicks both do better when the figure is alarming.
The real finding is not that AI erased hundreds of thousands of jobs last year. It is that the easiest way into professional work is narrowing, fastest in exactly the sector Ghana competes in. That is a slower problem than a layoff headline, and a harder one to plan around.
Sources
- Caribou and Genesis Analytics with the Mastercard Foundation, Preparing for AI in the BPO and ITES Sector in Africa, 3 April 2025. mastercardfdn.org
- US Bureau of Labor Statistics, paralegal and legal assistant employment projections 2025–2035.
- Business Day Ghana, 28 May 2026 — Ghana's position in the 2026 Global Outsourcing Talent Index. businessdayghana.com
- Legal Technology / RSGI reporting on Harvey adoption across Am Law 100 firms.
Common questions
Has AI actually eliminated large numbers of jobs?
Not in the way viral posts claim. US Bureau of Labor Statistics projections for paralegals — a commonly cited example — show roughly 0% change between 2025 and 2035, with about 40,800 openings a year. What the data does show is suppressed growth: jobs that would have been created are not being created.
What does the "40% of tasks" figure actually mean?
Research by Caribou and Genesis Analytics with the Mastercard Foundation projects that around 40% of tasks in Africa's outsourcing and IT-enabled services sector could be automated by 2030. It measures tasks, not jobs. A role loses some of its work rather than vanishing, though that usually means fewer people are needed.
Why are entry-level jobs most at risk?
Because junior work is the most repetitive. In Africa's outsourcing sector, entry-level roles are 68% of the workforce and more than half of their tasks are automatable. In finance and accounting, nearly two-thirds of junior tasks are exposed.
Is Ghana's outsourcing sector in trouble?
Ghana ranked 17th globally in the 2026 Global Outsourcing Talent Index, which is a strong position built on cost, talent and English proficiency. That same strength is what exposes the country to automation of routine outsourced work.
What skills are hardest to automate?
Work where a named person is accountable for the result — judgement calls, sign-off that carries professional or legal weight, and anything needing local presence, relationships or regulatory knowledge.