■ CRITICAL RISK ■ Science & Research
Yes — the compiling, tabulating, and checking that defined the role now happens in a script, and AI writes the script. The statistician kept the judgment; the assistant's tasks became a code library.
“R and Python don't need an assistant. They ARE the assistant.”
Our AI replacement risk score — how we score jobs
Statistical assistants do the labor beneath the analysis: compiling data from surveys and records, entering and cleaning it, computing means and rates and cross-tabulations, checking figures for errors, formatting tables and charts for reports, and maintaining the datasets a statistician or economist actually analyzes. In government statistical agencies, universities, and market research firms, it was the classic support tier — meticulous work that required numeracy but not a statistics degree.
Every layer of that stack has been automated in sequence. Survey data now arrives digitally, skipping entry. Data-cleaning pipelines flag outliers and inconsistencies programmatically. The tabulations that once filled an assistant's week are a few lines of R or Python that run in seconds and never transpose a digit. Charting libraries and reporting tools generate the tables and figures straight from the data. The newest layer cuts deepest: large language models now write the cleaning and analysis code from a plain-English request, which means the statistician doesn't even need to know the syntax — the historical reason an assistant existed at all. When the boss can type 'crosstab response by region and flag cells under 30' and get working code, the support tier's remaining rationale evaporates.
What resists sits at the edges of the data, not the math: understanding why a source's numbers look wrong, chasing down a reporting agency's definition change, judging whether messy real-world records are fit for purpose, and the survey-operations work of actually collecting data from humans. Those tasks bleed upward into data analyst and survey methodologist roles rather than remaining assistant work. Our 92 reflects a role whose job description reads like a scripting language's feature list — because that's what it became.
Automatability: our editorial assessment of current and near-term AI capability
Deep into the process and accelerating. Scripting already absorbed the tabulation work over the past two decades; AI code generation is now removing the last reason to staff a support tier, since analysts produce their own pipelines conversationally. Through this decade, expect remaining statistical assistant positions — mostly in government agencies with slow job-classification systems — to be reclassified as data analyst roles or quietly retired.
Barely, and shrinking. The compiling-and-tabulating tier the title describes has been absorbed by scripting, and AI code generation removed the last barrier — analysts now produce their own pipelines without support staff. The positions that survive are mostly in government classification systems that update slowly. The realistic path runs through the title, fast, toward data analyst.
It's collapsing the ladder's bottom rung. Data cleaning, tabulation, and chart production — the traditional apprenticeship tasks — now come out of a language model prompt in seconds. That's efficiency for organizations and a genuine problem for newcomers, because the route into statistical careers increasingly requires arriving with analyst-level skills rather than growing into them on the job.
Learn the combination: enough statistics to know when an analysis is wrong, enough programming to build it, and enough domain knowledge to know why it matters. Pure execution skills — running the numbers — are exactly what automated. Judgment about data quality, method choice, and interpretation is what employers still pay for, at analyst titles and above.
Treat your position as a funded classroom. Learn R or Python using your actual work as practice material, automate your own tasks before someone else does, and push for reclassification to analyst once you're doing analyst work. In parallel, master the AI coding tools — the durable role is the person who verifies and directs machine-produced analysis.