ISCO 3411 · O*NET 23-2011.00 · codes unverified
Paralegal
Highly exposed and highly assisted at the same time. The difficulty is that the exposed tasks are the ones juniors used to learn on.
One of the roles where the apprenticeship problem in chapter three bites hardest. Read the essay →
A large share of the work is moving. The remaining core is narrower and more senior than it used to be.
Scale, labels and thresholds are provisional. This is open question Q-02.
16 of 20 in the launch set
How the score is made
Four declared inputs, never collapsed into one number. Assistance is kept apart from outright automation, because a tool that makes you faster and a tool that replaces you are not the same event. Physical work counts as shelter only to the extent robotics cannot reach it.
The full method is written out, including what it cannot tell you.
Note on this occupation This page is the clearest case for the apprenticeship thread. The tasks being automated are the training ground.
- Assisted
- 74
- Done outright
- 68
- Physical
- 8
- Robotics reach
- 40
Higher is generally good news. It means more of the work gets faster without the work going away.
This is the number most people mean when they say exposure. It is kept separate from assistance on purpose.
Protective only for as long as robotics cannot reach it. Read this one next to the robotics number, never alone.
Low reach turns physical work into shelter. High reach turns it into the main exposure.
placeholder Working Paper 140, Generative AI and Jobs 2025 index O*NET 31.0 Gen AI Capacity Study September 2025 detailed data release data 0.1-placeholder
What changes
Tasks moving to machines, in whole or in part.
- Document review and first-pass discovery.
- Drafting standard correspondence and bundles.
- Citation checking and precedent searching.
What stays human
Tasks that need a person, and why.
- Noticing the document that does not fit the pattern.
- Managing a client who is frightened and under-informed.
- Carrying responsibility for an error a court will see.
What's new
Work that did not exist in this job ten years ago.
- Verifying machine-drafted work to an evidential standard.
- Running matters at volumes a firm could not previously staff.
- Explaining to a client what was automated and what was not.
These three lists are editorial commentary, written against task-level data but not generated from it (OCC-05). Treat them as the author's reading, not as output.
Compare with another job
Pick one. You get a card with both scores, both sets of components, and a link worth sending to someone.
- Hairdresser 84
- Plumber 82
- Primary school teacher 79
- Electrician 79
- Aged care worker 75
- General practitioner 73
- Registered nurse 72
- Chef 69
- Fashion designer 65
- Civil engineer 59
- Product designer 55
- Software developer 44
- Graphic designer 43
- Heavy truck driver 34
- Data analyst 32
- Accountant 28
- Copywriter 24
- Bank teller 13
- Customer service representative 12
Sources used on this page
- Working Paper 140, Generative AI and Jobs
International Labour Organization · 2025 index · Confirm
- O*NET
US Department of Labor · 31.0 · CC BY 4.0
- Gen AI Capacity Study
Jobs and Skills Australia · September 2025 detailed data release · Confirm