Free sampler. Re-scored every six months.
General document translation scores 8.8 out of 10 for AI exposure — the highest score of any career in this index, where 10 is most at risk. But court and medical interpreting scores 4.3. Same languages, same training, a 4.5-point gap. Translation is the profession where the automation has largely already happened, which makes it the most instructive career we score.
The short answer for parents: not as a route to translating documents. That work is effectively gone and has been for several years. Interpreting — a human physically present, legally accountable, in a room where it matters — is a different and genuinely protected career that happens to use the same languages.
Every career in this index is scored 1–10, where 10 is most exposed to AI. Same six factors, same weights, applied identically to an interpreter and a paramedic.
| Translation track | 2023 | 2025 | Now | 3-yr move | Band |
|---|---|---|---|---|---|
| Community / medical interpreting | 3.5 | 3.9 | 4.3 | +0.8 | Low–Mod |
| Legal / court interpreting | 3.6 | 3.9 | 4.3 | +0.7 | Low–Mod |
| Conference / simultaneous interpreting | 4.2 | 4.6 | 5.1 | +0.9 | Moderate |
| Literary translation | 5.9 | 6.5 | 7.0 | +1.1 | High |
| Localization / technical translation | 7.1 | 7.7 | 8.2 | +1.1 | High |
| General document translation | 7.7 | 8.4 | 8.8 | +1.1 | High |
2023 and 2025 figures are reconstructed using current methodology, not archived from past editions.
For scalethe median career in this edition scores around 5.5. Graphic design production scores 8.4. Entry-level software development scores 8.1. Bedside nursing scores 2.8.
Two things to notice. The 4.5-point internal spread is the widest of any profession we score. And document translation was already at 7.7 in 2023 — this automation did not begin with the current wave. It largely finished during it.
For most written translation, it substantially already has.
This is not a forecast. Machine translation crossed the threshold of commercial usability years ago, rates for general translation work collapsed accordingly, and the profession restructured around post-editing rather than translating.
How general document translation rates against each. Ratings are 0–10 on each factor's own terms.
So you can see what the analysis actually looks like.
How much of this job can AI already do? — rated 9.5
Every other high score in this index is a forecast about a process underway. Translation is not. It is a report on something that has already happened, which makes it the most useful case study we have.
Text-to-text conversion is the single task these systems were built for. Not adjacent to it, not an application of it — the thing itself. Take language in, produce equivalent language out. There is no profession more directly in the path of the technology, and the results are visible in the market rather than in projections: rates for general translation fell, agencies restructured around post-editing machine output, and the job title that survived is "post-editor" rather than "translator."
We rate it 9.5 rather than 10 because a human still reviews high-stakes output, and because some domains resist — legally binding text, marketing that must land culturally, anything where an error is expensive.
Now look at the other end of the same profession. Court interpreting rates 6.5 on this factor — still substantial, but far lower, and it scores 4.3 rather than 8.8. What changed?
The interpreter is physically present in a room (6.5 on physical, against 1.0). Someone is legally accountable for an accurate rendering, and in many jurisdictions certified (7.5 on regulatory, against 1.0). A defendant or patient is trusting a specific human at a moment that matters (8.0 on trust, against 2.5).
Same languages. Same skill. Entirely different job, because of where it happens and who is answerable for it.
The general lesson, and it is the single most transferable finding in this index: when a profession's core task is converting one form of text into another, expect no protection at all. When the same expertise is applied in a room, with a person, under accountability, expect a great deal. The question to ask of any career is not "what do you know?" but where do you have to be, and who answers for it?
Ranked by exposure, safest first:
Not as vocational training for translation. Almost certainly yes for other reasons.
The translation career specifically has been hollowed out, and honest programs should say so. A degree sold as preparation for translating documents is preparing students for the most automated work in this index.
But a languages degree was never only that. Fluency plus a domain remains genuinely valuable — international law, diplomacy, global health, trade, security, and any field where being able to operate in another culture rather than merely convert its text is the point. Interpreting is a real and protected career requiring specific additional training.
The distinction worth teaching a student: AI has largely solved translation and has not solved communication. Knowing what a sentence means is now free. Knowing what should be said, to whom, in what register, in a room where it matters, is not.
Worth asking any program: what proportion of graduates work in language-specific roles, and does the curriculum train interpreting and cultural competence or only written translation?
This sampler tells you where translation stands. The full profile tells you what to do about it.
Most families are weighing two or three careers seriously, and a few more they haven’t ruled out. Pick the ones you need.
Each includes a short version written directly to the student and the technical scoring appendix. Spring 2027 updates of whatever you buy are included.
Read one complete profile free → We publish computer science in full so you can judge the depth before buying anything.