Share and intensity of work current AI systems can materially affect.
SOC 15-2051
Data Scientists AI displacement risk
AI coding assistants accelerate model prototyping, feature engineering, and analysis code. Problem formulation, experiment design, model validation, and communicating uncertainty to decision-makers keep the role firmly augmentation-led.
Likely potential for exposed tasks to move to software after workflow integration.
Routine modeling work is being absorbed by automation and analyst self-service. Scientists who own ambiguous, high-stakes problems and production impact stay scarce.
Distribution
Where Data Scientists sits across 620 tracked roles
Displacement pressure 36 — higher than 61% of the 620 occupations tracked on displacement.ai.
Score version
This page uses Seed model v0.4 (seed-v0.4-2026-05), last reviewed 2026-08-08. Directional occupation-level planning model using hand-reviewed public research, task exposure estimates, wage context, and transition-pathway assumptions.
30 O*NET task statements matched to SOC 15-2051. The displayed task profile combines these official task statements with the current public score model.
Median wage context: $120,230 (May 2025, US national). The latest BLS row matched SOC 15-2051.
Scores are planning signals, not forecasts. Local hiring demand, employer-specific workflows, licensing, and credentials must be validated before making career decisions.
2030 economic stress test
How Anthropic's scenarios classify Data Scientists
SOC 15-2051 places this role in the paper's cognitive occupation group. These group-level outcomes do not change the 36/100 role score and are not an occupation forecast.
+0.4% group wage
-0.5% cognitive employment since mid-2026; 2.9% cognitive unemployment.
Economy-wide: +1.6% GDP and 3.9% unemployment.
-0.3% group wage
-3.9% cognitive employment since mid-2026; 4.5% cognitive unemployment.
Economy-wide: +8.3% GDP and 4.6% unemployment.
-11.5% group wage
-21.5% cognitive employment since mid-2026; 17.9% cognitive unemployment.
Economy-wide: +32.4% GDP and 11.9% unemployment.
Compare the assumptions and limitations across all three scenarios. Source: The Anthropic Institute Working Paper No. 2026-02.
Official task evidence
O*NET task matches for Data Scientists
The current evidence import matched 30 task statements from Task Statements 31.0 (August 2026). These rows are used as a grounding layer for judging which parts of the occupation are repeatable, language-heavy, analytical, social, physical, or compliance-sensitive.
Dataset31.0 (August 2026)
Matched tasks30
SOC15-2051
- Core task / ID 21823
Analyze, manipulate, or process large sets of data using statistical software.
- Core task / ID 21828
Create graphs, charts, or other visualizations to convey the results of data analysis using specialized software.
- Core task / ID 21837
Test, validate, and reformulate models to ensure accurate prediction of outcomes of interest.
- Core task / ID 21829
Deliver oral or written presentations of the results of mathematical modeling and data analysis to management or other end users.
- Core task / ID 21836
Recommend data-driven solutions to key stakeholders.
- Core task / ID 21831
Identify business problems or management objectives that can be addressed through data analysis.
Source: O*NET Resource Center, Task Statements. Raw import target: data/raw/onet/task-statements-31-0.txt.
Task profile
Where AI changes the work
Build and validate predictive models
Exposure 62, automation 32%, augmentation 76%.
O*NET evidence: Test, validate, and reformulate models to ensure accurate prediction of outcomes of int... (ID 21837)
Write analysis code
Exposure 74, automation 40%, augmentation 74%.
Clean and process large data sets
Exposure 78, automation 54%, augmentation 62%.
O*NET evidence: Analyze, manipulate, or process large sets of data using statistical software. (ID 21823)
Present findings to stakeholders
Exposure 36, automation 10%, augmentation 52%.
O*NET evidence: Recommend data-driven solutions to key stakeholders. (ID 21836)
TaskExposureAutomationAugmentation
Build and validate predictive models6232%76%
Write analysis code7440%74%
Clean and process large data sets7854%62%
Present findings to stakeholders3610%52%
Transition pathways
Adjacent moves that preserve existing skills
adjacent role
ModerateMachine Learning Engineer
Training horizon: 4-9 months. Skill overlap 72. Wage preservation signal 112.
- Deploy models to production
- Learn MLOps tooling
- Own monitoring and retraining
role redesign
ModerateApplied AI Scientist
Training horizon: 3-6 months. Skill overlap 78. Wage preservation signal 118.
- Evaluate foundation models
- Design domain evaluations
- Prototype retrieval systems
Comparison guides
Compare the next move before you commit
4-9 months
Data Scientists to Machine Learning Engineer
Compare AI displacement pressure, wage preservation, skill overlap, training time, and first proof project for moving from Data Scientists into Machine Learning Engineer.
3-6 months
Data Scientists to Applied AI Scientist
Compare AI displacement pressure, wage preservation, skill overlap, training time, and first proof project for moving from Data Scientists into Applied AI Scientist.
What the AI risk score means for Data Scientists
The displacement pressure score for Data Scientists is 36. That score blends task exposure, automation pressure, augmentation potential, wage vulnerability, transition feasibility, and source confidence. It is designed to help workers and workforce teams decide where to act first, not to claim a specific date when a job will disappear.
For this role, the clearest risk pattern is visible at the task level. Clean and process large data sets carries 54% automation pressure, while Build and validate predictive models carries 76% augmentation potential. That means the best response is usually a targeted redesign of work: move away from repeatable production tasks and toward judgment, exception handling, coordination, stakeholder context, and accountable use of AI tools.
Labor-market context and wage risk
Median wage: $120,230 (May 2025, US national). Employment context: Fast-growing technical role with strong wage buffer. Typical education: Bachelor's degree; advanced degrees common.
Wage vulnerability is 20, while transition feasibility is 80. A high wage-vulnerability score means workers should pay close attention to salary preservation before making a move. A high transition-feasibility score means there are adjacent paths that can reuse existing skills without requiring a complete career reset.
- Low to moderate displacement pressure
- High augmentation upside
- Demand favors business-aware scientists
Upskilling priorities
Skills that make this role more resilient
The safest upskilling plan starts with skills already close to the work. For Data Scientists, the strongest near-term skill priorities are listed below. These are useful whether the goal is to stay in the role, move to a redesigned version of the role, or transition into an adjacent occupation.
Experiment design
Build proof of this skill through a work sample, checklist, dashboard, case note, workflow map, or portfolio artifact tied to the transition paths on this page.
Model validation
Build proof of this skill through a work sample, checklist, dashboard, case note, workflow map, or portfolio artifact tied to the transition paths on this page.
ML system literacy
Build proof of this skill through a work sample, checklist, dashboard, case note, workflow map, or portfolio artifact tied to the transition paths on this page.
Decision communication
Build proof of this skill through a work sample, checklist, dashboard, case note, workflow map, or portfolio artifact tied to the transition paths on this page.
90-day transition plan
The most practical next step is not to wait for a layoff or a full role redesign. Use the next 90 days to create evidence that you can operate in a safer, more AI-augmented version of the work.
- In the first 30 days, document the repetitive tasks in your current work and identify where AI can reduce drafting, lookup, classification, or reporting time.
- By 60 days, complete one small project connected to Machine Learning Engineer, such as deploy models to production.
- By 90 days, compare internal openings and external postings for Machine Learning Engineer or Applied AI Scientist and update your resume around measurable workflow outcomes.
FAQ
Questions about AI and Data Scientists
Will AI replace Data Scientists?
AI coding assistants accelerate model prototyping, feature engineering, and analysis code. Problem formulation, experiment design, model validation, and communicating uncertainty to decision-makers keep the role firmly augmentation-led. The better planning signal is not full replacement, but which tasks become automated, which tasks become AI-assisted, and which responsibilities still need human judgment.
Which parts of Data Scientists work are most exposed to AI?
Clean and process large data sets and Write analysis code show the strongest automation pressure in this model. Build and validate predictive models and Write analysis code are better treated as AI-augmented work.
What should Data Scientists learn next?
Start with Experiment design, Model validation, ML system literacy. The most practical adjacent paths in this model are Machine Learning Engineer and Applied AI Scientist.
How should this score be used?
Use it as a planning signal, not a prediction. Confirm local hiring demand, wages, licensing, credentials, and employer adoption before making a career move.
Sources
Evidence trail
International Labour Organization / 20 May 2025Generative AI and Jobs: A Refined Global Index of Occupational ExposureMcKinsey Global Institute / 26 July 2023Generative AI and the Future of Work in AmericaOECD / 11 July 2023OECD Employment Outlook 2023: Artificial Intelligence and the Labour MarketO*NET Resource Center / Database 31.0 (August 2026)Occupation, Task, Skill, and Work-Activity DataUS Bureau of Labor Statistics / May 2025 estimates (released 15 May 2026)Occupational Employment and Wage Statistics