Is AI Coming for Your Job? Here’s How to Tell
For a long time, I have repeatedly written about the significant benefits of properly built predictive financial models. Indeed, you could say that I have been an ardent advocate of financial models for most of my adult life, since I started when I was 35 years old.
Obviously, we are all aware of the existence of AI and the rapid rate at which it is developing.
I have been collaborating with AI for the last year or two, and have experienced its dramatic growth in capability from a large language scraper to its current ability as demonstrated in Open AI – ChatGPT4 – o3.
In truth it’s been a two-way street. I have been teaching ChatGPT about building a comprehensive financial model, incorporating the FAST standard principles, and it, in turn has been invaluable in providing me with relevant data for the building of complex models – e.g. the input assumptions for a proposed new multi disciplinary state hospital, just to name one. We are also collaborating on Goalfix’s first SaaS modelling application – a start-up model for aspiring entrepreneurs.
When we started our relationship, our dialogue was a bit formal, but after a number of interactions, I gave him a name – R-CHI , pronounced Archie. I also identified myself as Colin.
Today, our conversations and collaborative work is akin to two good friends.
In fact, I might say, that today Archie is my best modelling buddy!
So, one could say that I am acutely aware of AI.
I recently came across a very interesting – and somewhat scary – article and I felt it would be useful to share it.
I trust that you will find it useful.
Just a point – there is still a question mark regarding ChatGPT’s ability for creative thought!
Kind regards,
Colin Human CA(SA)
Title: Is AI coming for your Job
A line of job seekers looking for jobs that may soon no longer exist with the growth of generative AI.
Back in 2023, Goldman Sachs warned that generative AI could put 300 million jobs at risk worldwide.
By 2025, experts warn that AI could wipe out half of all entry-level, white-collar jobs—and spike unemployment to 10%-20% over the next several years.
Large language models (LLMs) like Claude or ChatGPT can now write marketing copy, compose poetry and short stories, draft legal memos, and debug code in seconds.
It can search the web, collate sources, generate research summaries, and even spit out polished slide decks.
That makes many people wonder: Is my job next?
Recent research suggests the answer could depend less on your job title and more on the bundle of tasks you perform each day.
Think of tasks as the sub-units of work that fill your calendar: drafting an invoice, negotiating with a supplier, sketching a storyboard frame, reconciling a ledger entry, or writing some code.
Depending on how any of these tasks can be automated with AI, you might or might not start to worry.
Below, we explain how to gauge your risk and potential upside amid the AI rollout.
KEY TAKEAWAYS
- The more of your daily tasks that large-language models (LLMs) can already handle, the higher your displacement risk.
- Workers whose task mix ranges from easily automated to hard-to-automate will likely fare better than specialists who do one thing well.
- With the right design and policy, the technology could revive middle-skill, middle-income work rather than destroy it.
Task Exposure: The Metric to Watch
No surprise here: jobs composed mainly of tasks that AI can do entirely are most at risk.
On the other hand, those that involve at least some human-only tasks appear to be safe (for now), as employees shift to the creative, client-facing, uniquely human tasks that AI still can’t do.
Run a mini-audit on yourself: list your top 10 weekly tasks and tick off any that a GPT-4-level model could do today.
If AI could handle more than 50%, that signals displacement risk; under 30% suggests that AI could provide productive augmentation.
| Example Tasks-at-Risk | |
|---|---|
| Task | Likelihood an LLM Can Do It Today |
| Draft a marketing email announcing a new product | High |
| Translate a memo from English to Spanish | High |
| Summarize a 20-page research article into five bullet points | High |
| Proofread an article or blog post for grammar and style | High |
| Generate a first-pass legal memorandum citing precedent | Moderate |
| Build a financial model with bespoke tax rules in Excel | Moderate |
| Analyze customer sentiment from 100 call transcripts and flag hot issues | Moderate |
| Write a song or compose music | Moderate |
| Negotiate contract terms with a long-standing client over a Zoom call | Low |
| Troubleshoot a noisy car engine in the shop | Low |
| Facilitate an in-person brainstorming session for a fresh ad concept | Low |
History Says Disruption Arrives in Waves—Not Overnight
If we look to history, we find that technological disruption tends to diffuse through the labour market over a period of years.
Indeed, the U.S. job market actually changed more slowly from 1990-2017 than in any earlier period, despite the arrival of computers and the internet.
For career planning, that means AI shock is unlikely to hit all at once like a meteor; instead, watch for gradual but compounding shifts.
Workers who track such early indicators can pivot before the crest of the wave, much like typists who re-skilled into desktop-publishing roles during the early days of the personal computer.
We are already seeing some strong signals: sharp declines in retail jobs, stalled growth in low-paid services, rapid STEM hiring, and shrinking middle-wage employment—all of which might indicate the pace has begun to accelerate.
Author: ADAM HAYES
Updated June 09, 2025
