EXPOSED
Written translation — documents, subtitles, localization strings, marketing copy — is the single most thoroughly automated language task there is, and the market has already repriced it toward machine-translation post-editing at lower rates per word. Live interpreting holds up better: courtroom and hospital work demands physical or on-call presence, sworn accuracy, cultural mediation, and a certified human whose name goes on the record. The modal worker in this SOC straddles both, which is why the score lands on the COOKED/EXPOSED line rather than clearly above it.
This fall is concentrated in 2020 and has not recovered since.
Median pay $51,830 → $60,170 -7.1% in real terms
This line is counted by the Bureau of Labor Statistics — the one figure on this page that isn't a judgement of ours. Headcount moves on demand, offshoring, demographics and the business cycle, and automation is one term among several, often not the loudest.
So a falling line is not evidence that AI did it, and a rising one is not evidence that it won't. Both happen in this register: some occupations resist automation and shrink anyway, others are highly automatable and keep growing. The marked year is 2020.
BLS projection, 2024–2034
+1.7%
Percentage only. The projection counts a different population from the 52,060 above — it includes self-employed workers, which for this occupation is most of them, so the two headcounts are not comparable.
Exposed, but growing
AI can already do a lot of these tasks, and the BLS still expects +1.7% more of these jobs by 2034. Demand for the output is growing faster than the work is being automated away — the mechanism BLS gives for software developers, and the combination people most often misread as an error.
Different clocks. The score is what current AI could do to this work today. The projection is how many of these jobs will exist in 2034. Everything between the two — how fast employers actually adopt, whether demand grows in the meantime — is why they can point opposite ways without either being wrong.
~6,900 openings a year on average, including replacing people who leave.
LinguistTranslatorInterpreterFarsi LinguistDeaf InterpreterLegal TranslatorRussian LinguistArabic TranslatorCourt InterpreterBraille TranslatorEnglish TranslatorSpanish TranslatorBilingual SecretaryBraille TranscriberContract TranslatorLanguage TranslatorMedical InterpreterSpanish InterpreterCryptologic LinguistFreelance TranslatorLanguage InterpreterTechnical TranslatorBilingual InterpreterCommunity Interpreter
Holding it up: trust premium . Weakest point: task resistance .
Core tasks are already automatable A 5 reflects that the bulk of billable volume in this SOC — contract translation, subtitle files, software strings, birth certificates, patent abstracts — now arrives as an MT draft to be post-edited, and even simultaneous interpreting has working speech-to-speech pipelines in conference and telehealth queues; the score is not 2 only because whispered courtroom simultaneous, ASL, and low-resource language pairs still require a live human ear.
Some physical or field component A 7 covers the interpreter who drives to the deposition, stands next to the patient during a pelvic exam, sits in the booth with a headset, or does relay work over VRI carts — real presence and travel, but nothing manipulated with the hands and no uncontrolled site conditions beyond a noisy courtroom or an ambulance bay, which is why it sits below the 13 line.
Certification preferred, not legally required Court certification (state or Administrative Office of the Courts) and CCHI/NBCMI medical credentials are gatekeepers for the good assignments and you swear an oath on the record, but no statute reserves the act of translating to a licensed person — an unsertified bilingual can and does take the work, and ATA certification is a marketing asset rather than a legal monopoly, which puts this at 6 rather than 12.
Meaningful discretion A 7 reflects real calls made in the moment — flagging a register mismatch, deciding whether an idiom or a hedged consent phrase carries over, interrupting to request clarification on the record — bounded by canons of ethics that mandate completeness and prohibit editorializing, so you own accuracy but not the underlying decision, unlike the physician or judge in the room.
The verdict above describes this occupation as a whole. Almost nobody does the typical version of a job — tick what's actually in your week and see how your own mix sits.
Your task mix speaks to task resistance (5/20 here) — how much of the day's work current AI already does. That is the dimension the boxes above are about.
It cannot move the other three. Liability shield (6/20) is whether the law requires a licensed human to sign. Trust premium (9/20) is whether buyers specifically pay for a person. Judgment and accountability (7/20) is whether the role exists to own consequential calls. Those are facts about the occupation's standing, not about which tasks are in your week — a paralegal who does only trial exhibits still holds no licence. Together they are 22 of this occupation's 34 points (65%).
Embodiment (7/20) is also a property of the work rather than the worker, but we don't tag individual tasks as physical or not, so the picker can't tell you anything about it. That's a limit of this tool, not a claim.
Did we get the list right? Tell us what's missing — the tasks are written from the outside, and you're reading this from the inside.
The moves above are yours to make. This is the other half: what would have to change in the world for the occupation itself to score higher. None of it is in any one person's gift, but it is where the floor actually comes from. Scores here are not a one-way ratchet. Only two of the five dimensions — task resistance and embodiment — track what machines can do. The other three track law, what buyers will pay for, and who is answerable, and those move in both directions, often in response to the same pressure AI creates. If every lever below landed, this occupation would score around 52/100, still EXPOSED.
Federal or state court rules explicitly barring machine translation for record proceedings and requiring a credentialed interpreter's oath — e.g. state judiciary language-access plans (California, New Jersey, Washington) codifying that only a court-certified interpreter may render testimony, plus USCIS/asylum and Title VI hospital rules requiring a qualified human interpreter's attestation rather than an AI relay. Similar for sworn/certified translations for immigration, adoption, and patent filings where a translator's signed certificate of accuracy carries perjury exposure.
Task-mix shift as raw-draft translation disappears entirely: what remains is legally-consequential certified translation, simultaneous conference and court interpreting under time pressure, ASL and low-resource-language work with thin training data, and transcreation where the deliverable is a judgment call about register and cultural risk rather than a rendering. This is a genuine two-tier occupation and the bottom tier is already largely gone.
Malpractice insurers and hospital accreditation bodies (Joint Commission language-access standards) requiring documented human interpreter involvement for consent, discharge, and psychiatric encounters after an adverse-event case traced to machine translation.
Formalization of the interpreter's affirmative duty to intervene — codes of ethics (NCIHC, RID, court interpreter codes) already require flagging miscommunication, cultural framing, and interpreter conflict; if court rules or hospital policy make the interpreter the named party responsible for halting a proceeding or encounter when comprehension fails, the role owns a consequential call.
Buyer-side backlash in specific niches: literary publishers and author contracts prohibiting MT (Authors Guild model clauses, translator credit-on-cover campaigns), and audiovisual union agreements on subtitle quality after high-profile bad-subtitle incidents on streaming platforms. Narrow, prestige-segment only — it does not reach the commercial document market.
On-site presence requirements for court, deposition, medical, and conference work if remote-interpreting platform standards tighten (some judiciaries restrict video remote interpreting for jury trials). Marginal — the trend runs the other way toward remote.
The limit. The ceiling is structural: the certified-interpreter tier that carries the liability and judgment is a minority of this SOC's headcount, and gains there do not lift the written-translation majority. Realistic top is mid-50s for the courtroom/medical interpreter, while document translators stay near the current score regardless of what happens to interpreting rules.
| Houston-Pasadena-The Woodlands, TX | 2,530 | $44,490 -26% |
| New York-Newark-Jersey City, NY-NJ | 2,520 | $87,590 +46% |
| Phoenix-Mesa-Chandler, AZ | 1,720 | $50,070 -17% |
| Boston-Cambridge-Newton, MA-NH | 1,660 | $75,120 +25% |
| Washington-Arlington-Alexandria, DC-VA-MD-WV | 1,630 | $94,510 +57% |
| Los Angeles-Long Beach-Anaheim, CA | 1,480 | $69,460 +15% |
| Dallas-Fort Worth-Arlington, TX | 1,280 | $56,920 -5% |
| Tampa-St. Petersburg-Clearwater, FL | 1,170 | $49,470 -18% |
| San Jose-Sunnyvale-Santa Clara, CA | 300 | $101,530 +69% |
| Washington-Arlington-Alexandria, DC-VA-MD-WV | 1,630 | $94,510 +57% |
| Bridgeport-Stamford-Danbury, CT | 60 | $93,960 +56% |
DeepL · Warhorse Studios · Duolingo · State of New Jersey
Startup Fortune reports that translation company DeepL cut 25 percent of its staff, attributed to competitive pressure from general-purpose AI models.
AI translation firm DeepL reportedly cut around 250 jobs as part of a restructuring toward an "AI-native" organization.
Tech.eu reports that German AI translation firm DeepL will cut 250 staff.
The Economic Times reported that translation company DeepL cut about a quarter of its staff, attributing the reduction to AI.
An English translator for Kingdom Come Deliverance 2 says Warhorse Studios replaced his role with AI translation to cut costs, telling him the position was obsolete.
Amazon announced Kindle Translate, an AI-powered translation service enabling Kindle Direct Publishing authors to produce eBooks in other languages without commissioning human translators.
Has AI actually changed your work? One tap, anonymous, and the running tally is public. Nothing else is asked of you.
Rather than check back: get the digest and we'll tell you what changed — or watch a single occupation from its own page.