LO
Interpreter — tasks, one by one
The unit of analysis is the task, not the job title. Each one below carries its direction, whether the judgement rests on evidence or on platform inference, the reasoning, and what it does not establish.
Every task on this page#
Court and legal interpreting
Still human-led✓ Evidence-backedInterpreting for defendants, witnesses and parties in courts and police proceedings.
The law requires a qualified person. In US federal courts, the judge shall use the most available certified interpreter, in simultaneous mode for parties to proceedings brought by the United States. EU law requires interpretation in criminal proceedings to be of a quality sufficient to safeguard the fairness of the proceedings. A US guidance body for state courts says courts should never use machine translation for court events or to convey legal or procedural information.
US federal law, one EU directive and guidance for US state courts; other countries differ, and none of this measures interpreters' hours.
Medical interpreting
Still human-led≈ Platform inferenceInterpreting between patients and clinicians, in person or remotely.
US health rules require a qualified interpreter, and machines still make serious errors here. Under the 2024 Section 1557 rule, a covered entity must offer a qualified interpreter when interpretation is required, and machine translation of critical text must be reviewed by a qualified human translator. In one study of real clinic speech, Google Translate's voice mode had linguistic or clinical errors in 33.3% of segments against 4.8% for qualified interpreters; in another, none of three apps met the non-inferiority threshold against professional interpreters for simple clinical exchanges.
US rules and two small studies of specific apps and languages; newer tools may perform differently, and neither measures interpreter employment.
Conference and simultaneous interpreting
Being augmented≈ Platform inferenceInterpreting speeches and meetings in real time from a booth or remotely.
Machines now do this in the lab, and interpreters use AI around it. Comparing machine and human simultaneous interpreting, researchers found human interpreters better on intelligibility and the machine slightly better on informativeness. A technology company reports its own system reaching 81.3% and 78.0% on its own quality metric for Chinese–English interpreting. The European Commission tells its interpreters AI tools can help prepare interpretation work and interpreting in the booth.
Small comparisons, one with an author from an interpreting-platform company, and a company's own metric; not a measure of conference work moving to machines.
Sign language interpreting
Still human-led✓ Evidence-backedInterpreting between spoken language and sign language, in person or by video.
US disability rules define a qualified interpreter as one who interprets via video remote interpreting or on site, and set video standards around the interpreter's face, arms, hands and fingers. US projections expect demand for American Sign Language interpreters to grow with the increasing use of video relay services.
US rules and a projection; they do not measure sign-language interpreters' work or the state of sign-language technology.
Preparation and terminology
Being augmented≈ Platform inferenceResearching subjects and building glossaries before an assignment.
This is where interpreters use AI most openly. The European Commission's interpreting service says AI tools can help prepare interpretation work, warns about sensitive information, and points its interpreters to the Commission's own translation and speech-to-text tools.
One employer's guidance to its own interpreters; it does not measure how much preparation time changed.