Occupation score change
IMPACT 10paralegal exposure increased 1.0 points
AI Exposure Score moved from a previous calculation to 68/100 under methodology v1.0. 6 evidence item(s) contributed.
68/100
increased by +1.0 vs last
High exposure
3 Oct 2026
19 hours ago · sources: BLS OES (approx) · O*NET
6 sources
6 core tasks shown · 6 secondary hidden
Paralegals and Legal Assistants Very High exposure: 3 of 6 core tasks show high task automation potential : exposure 68/100 (high exposure).
No model has published evidence mapped to this occupation yet - the score comes from generic task assessment.
Score breakdown
Why this score?
Every component is stored with the score row and can be recomputed from the evidence table.
AI Exposure (composite)68/100
Methodology 2026-09-a · score v1.0 · input hash d95e93b1ce74350a
AI capability exposure79/100
Strength of current models on the capabilities this job's tasks require.
Task automation potential81/100
Task-level exposure, weighted by task importance. A component of the headline, not the headline itself.
Observed AI adoption62/100
Evidence that AI is actually being used for this work.
Labour-market signal0/100
Observed changes in postings, employment or earnings.
Evidence strength75/100
Quantity and source quality of the evidence behind the score.
6 evidence item(s) contribute. Calculated 19 hours ago. 3 Oct 2026 UTC.
Overview
Assist lawyers by investigating facts, preparing legal documents, or researching legal precedent. Conduct research to support a legal proceeding, to formulate a defense, or to initiate legal action.
Employment
≈350,000
approx. BLS OES
Median wage
≈$60,000
approx. BLS OES
Tasks tracked
12
linked to this occupation
Employment and wage values are approximate reference figures from the BLS Occupational Employment and Wage Statistics programme, stored with an approximation flag. Verify against the O*NET record and BLS before quoting them.
Tasks
Task-level automation exposure
Core tasks only. Exposure describes how much of the task output current AI can produce or substantially assist - not whether the task disappears. Secondary shared-baseline edges are hidden until per-occupation curation is complete.
core task · importance 4.4/5
core task · importance 4.3/5
core task · importance 4.1/5
core task · importance 4/5
core task · importance 3.9/5
core task · importance 3.9/5
Evidence
Evidence trail
6 attributed item(s). Confidence reflects study design: randomised and production deployments score higher than platform observations, which score higher than projections.
- task automationincreasesconfidence 92/100 · magnitude 8213 Jul 2023
In a randomised experiment on professional writing tasks, ChatGPT reduced time taken by about 40% and raised graded output quality by about 18% relative to the control group.
Direct experimental evidence for text-production tasks specifically. It supports a high automation component for routine writing, and says nothing about jobs that also involve judgement, sourcing or accountability.
Source: Noy & Zhang — Experimental evidence on the productivity effects of generative AI (Science) · Updated 13 Jul 2023
- researchincreasesconfidence 82/100 · magnitude 7424 Aug 2023
Using O*NET task descriptions scored by GPT-4, the authors estimate that about 80% of the US workforce could have at least 10% of their tasks affected by large language models, and about 19% could see at least 50% of their tasks affected.
This is a *capability and exposure* estimate produced by having a model rate task descriptions — not a measurement of employment change. It is the most-cited exposure figure in the field and is included here as an exposure input, weighted accordingly.
Source: Eloundou, Manning, Mishkin & Rock — GPTs are GPTs (arXiv:2303.10130) · Updated 24 Aug 2023
- researchdecreasesconfidence 78/100 · magnitude 4513 May 2024
Estimating that only a small share of tasks are currently cost-effective to automate, the paper projects total-factor-productivity gains of no more than about 0.66% over ten years, with limited aggregate labour-demand effects.
Included as a counterweight. Exposure — even high task exposure — does not imply large aggregate job loss, because adoption depends on cost, reliability, regulation and complementary investment. This evidence lowers the labour-market component rather than capability.
Source: Acemoglu — The Simple Macroeconomics of AI (NBER Working Paper 32487) · Updated 13 May 2024
- adoptiondecreasesconfidence 72/100 · magnitude 421 Feb 2024
The authors argue that AI can enable workers with less formal training to perform expert tasks, potentially rebuilding middle-class occupational demand rather than only displacing it.
A credible counter-direction hypothesis. Included so that the index does not systematically overstate substitution, and because the strongest experimental evidence (support agents, consultants) shows exactly this experience-compression effect.
Source: Autor, Levy & Solan — Applying AI to Rebuild Middle-Class Jobs (NBER 32140) · Updated 1 Feb 2024
- adoptionincreasesconfidence 70/100 · magnitude 6827 Feb 2025
Aggregated and anonymised Claude conversation data mapped to O*NET tasks shows that usage is concentrated in a small number of occupations and tasks — notably software development, writing and technical analysis — and that a substantial share of usage is augmentation rather than full automation of a task.
This is the primary *observed adoption* input, and it is deliberately kept separate from capability. A model being able to do a task does not mean firms are using it. Usage data is self-selected (Claude users) and therefore over-represents digital and technical occupations.
Source: Anthropic Economic Index · Updated 27 Feb 2025
- researchincreasesconfidence 55/100 · magnitude 4826 Mar 2023
The analysis estimates that a substantial share of current work tasks could be automated by generative AI, with a headline estimate of around 300 million full-time jobs globally exposed, and finds that most exposed occupations are partially rather than fully exposed.
Widely quoted and widely misread. It is a task-exposure scenario, not a forecast of unemployment. Used here at low-to-medium weight, and only for the exposure discussion, not for occupation-level scoring.
Source: Goldman Sachs Global Investment Research — The Potentially Large Effects of AI on Economic Growth · Updated 26 Mar 2023
Historical trend
Exposure over time
Backcast series: recomputed at quarterly checkpoints using only the models and evidence published by each date.
Reconstruction, not measurement. 5 observed daily snapshot(s) exist for this occupation; the earlier part of the curve is derived by applying the current methodology to the model releases and published evidence available at each past date.
Recent events
Signals affecting this occupation
- Research publishedIMPACT 40
Legal Research Bench: Measuring End-to-End Reliability in Long-Horizon Legal Research Agents
arXiv:2610.00609v1 Announce Type: new Abstract: Legal research is a core and time-consuming legal workflow. Lawyers must identify controlling authority, verify that it remains valid, reconcile statutes and cases, and synthesize a grounded answer. Language model agents are a natural fit for this retrieval-intensive workflow, and automating even part of it would be valuable. But that value depends on reliability: a single missing authority, stale citation, or wrong legal conclusion can make an otherwise plausible answer unusable. We introduce \textbf{Legal Research Bench} (LRB), a benchmark of 4
- AI adoption signalIMPACT 68
Measured AI usage concentrated in software development, writing and analysis
Aggregated and anonymised Claude conversation data mapped to O*NET tasks shows that usage is concentrated in a small number of occupations and tasks — notably software development, writing and technical analysis — and that a substantial share of usage is augmentation rather than full automation of a task.
- Research publishedIMPACT 45
Macroeconomic gains from AI are likely modest over a decade
Estimating that only a small share of tasks are currently cost-effective to automate, the paper projects total-factor-productivity gains of no more than about 0.66% over ten years, with limited aggregate labour-demand effects.
- AI adoption signalIMPACT 42
AI may expand expert work rather than only substitute for it
The authors argue that AI can enable workers with less formal training to perform expert tasks, potentially rebuilding middle-class occupational demand rather than only displacing it.
- Research publishedIMPACT 74
LLMs affect at least 10% of tasks for about 80% of US workers
Using O*NET task descriptions scored by GPT-4, the authors estimate that about 80% of the US workforce could have at least 10% of their tasks affected by large language models, and about 19% could see at least 50% of their tasks affected.
- Research publishedIMPACT 82
Writing tasks completed 40% faster with 18% higher quality ratings
In a randomised experiment on professional writing tasks, ChatGPT reduced time taken by about 40% and raised graded output quality by about 18% relative to the control group.
- Research publishedIMPACT 48
Estimate: about 300 million full-time jobs exposed to automation from generative AI
The analysis estimates that a substantial share of current work tasks could be automated by generative AI, with a headline estimate of around 300 million full-time jobs globally exposed, and finds that most exposed occupations are partially rather than fully exposed.
Related
Related Occupations
Occupations sharing major SOC group (23) with Paralegals and Legal Assistants.
faq
Frequently Asked Questions
Common questions regarding Paralegals and Legal Assistants and AI exposure.
Will AI replace Paralegals and Legal Assistants?
Current AI models show an exposure score of 68/100 for Paralegals and Legal Assistants. This measures technical capability to assist or automate specific tasks rather than predicted immediate job loss. High exposure indicates strong overlap between current LLM/multimodal capabilities and routine tasks in this occupation.
Which tasks in Paralegals and Legal Assistants have the highest AI exposure?
The most exposed tasks for Paralegals and Legal Assistants include "Investigate facts and law of cases and search pertinent sources, such as public records and internet sources, to determ…" (78% exposure) and "Gather and analyze research data, such as statutes, decisions, and legal articles, codes, and documents." (75% exposure).
What aspects of Paralegals and Legal Assistants can AI not do yet?
Tasks with the lowest AI exposure for Paralegals and Legal Assistants include "Meet with clients and other professionals to discuss details of cases." (5% exposure) and "Prepare for trial by performing tasks such as organizing exhibits." (20% exposure). These responsibilities typically require physical dexterity, licensed human accountability, or complex interpersonal navigation.
Widget
Embed this interactive AI exposure card into your article, research, or dashboard.
Track AI exposure changes for Paralegals and Legal Assistants
We monitor frontier AI model releases and benchmark updates. Enter your email for alerts when this occupation's exposure changes. No spam, ever.
By submitting, you consent to receive career disruption updates. We never share your data. Read our Privacy Policy.
Notice: All ratings shown are AI-estimated technical exposure scores measuring technical capabilities across O*NET task statements, not predicted job loss. Contains information from O*NET 31.0 by the U.S. Department of Labor, ETA, used under CC BY 4.0.