Is AI coming for me? › IT & software
Is AI coming for me?
Data scientist
Also called: Machine learning engineer, AI engineer
Anyone can ask a chatbot for a regression. Fewer know if it's nonsense.
High risk
50% of the work AI can already do
When it hits: 1–2 years
Big changes coming
LLMs and AutoML now handle exploratory analysis, standard modelling, charts and write-ups that used to justify a data science team. Business users increasingly do their own analysis through AI tools. The value left is in framing the right question, causal reasoning and knowing when the output is wrong.
What AI will take
- Exploratory data analysis and visualisation
- Building and tuning standard predictive models
- Writing Python and SQL for analysis
- Drafting reports and slide summaries of findings
- Feature engineering for common problem types
What stays human
- Framing a vague business problem into something answerable
- Designing experiments and causal analysis
- Spotting bias, leakage and misleading results
- Convincing sceptical decision-makers to act on findings
- Owning model risk for regulated decisions
How to adapt and stay ahead
- Get strong in causal inference and experiment design, not just modelling
- Move toward AI evaluation and model risk work, which regulators increasingly expect
- Get closer to the business so you're trusted to make decisions, not just produce charts