The report highlights that this type of MT use has so far tended to go under the radar, with little public discussion and what it deems to be scant evidence of the extent and nature of the reliance on MT tools in these contexts.
The data was collected between February and April 2024, using a database of pre-registered individuals. For this report, the profiles of survey respondents were filtered down to “police,” “medical/healthcare,” “health care and social assistance,” “emergency service,” and “legal services” workers.
Over a Third of UK Frontline Workers Use Raw MT
The research sample consisted of the survey responses of 2,520 UK professionals, 33% of whom reported using MT at work, most often in scenarios involving direct communication with others in a shared physical space, on personal devices, and through public browser interfaces, raising red flags about the protection of sensitive information.
For 72% of respondents, machine translation had never been mentioned in workplace training, and 15% reported that machine translation use was recommended by their employer.
Respondents expressed high levels of satisfaction with the tools they used and were generally confident in their ability to use the tools successfully, even though MT was not typically included in workplace training. Most expressed no concern about the resulting translation, despite their inability to determine levels of accuracy.
While Google Translate was the most often used tool, newer AI tools like ChatGPT were also used.
“We must also address the potential for AI to create a false sense of linguistic competence. The language industry’s complexities are already poorly understood by the general public and by frontline workers, and the advent of seemingly capable AI translation tools risks further obscuring the vital importance of human linguistic expertise. This misconception could lead to a further devaluing of language skills, ultimately impoverishing the UK’s linguistic capabilities.” — Uses of AI Translation in UK Public Service Contexts – Preliminary report foreword
Most respondents were native speakers of English. French was the second most spoken language among participants, and the language pair most often used for raw MT was English <> Polish. Other languages documented in the survey in order of frequency are Romanian, Arabic, Spanish, Urdu, Punjabi, and German.
The main reported reason for using MT was needing “to communicate with someone out loud in the same physical space.” Other reasons were reading or understanding something, exchanging written messages with someone via chat or similar text-based communications, and phone or online conversations.
The report also mentions that MT is not the only approach used by frontline workers to communicate in other languages. Respondents also cited getting interpreters, using AI, and resorting to pictures, charts, gestures, and signs.
2024 Slator Pro Guide: Translation AI
The 2024 Slator Pro Guide presents 20 new and impactful ways that LLMs can be used to enhance translation workflows.
NHS Guidance Advises Against MT
The vast majority (over 80%) of valid responses came from workers in the health and social care categories. According to the report, this participation reflects the overall demographics represented in the database and the country: the UK’s National Health Service (NHS) is the largest UK employer, and most respondents are healthcare workers.
Of note, existing NHS guidance advises against the use of MT technology. The report mentions that the results obtained in the study “may well be closer to the reality on the ground” and “more closely aligned with what they [the workers] saw as common practice than with official policies.”
Nurses, social workers, medical doctors, police officers, and attorneys were the most common professions among respondents. Legal and emergency services were the other groups represented.
The report also has some recommendations for institutions, including that they should formally acknowledge that staff will be inclined to use MT and that the use of AI translation should be addressed as part of policies flexible enough “to keep up with technological developments while also protecting the community from the risks posed by machine translation.”
Dom Hebblethwaite, CIOL’s Head of Membership & Ventures, said about the survey results that “if facts are misrepresented or key messages are mangled, public services quite simply fail the publics they serve. It is clear that the current situation of unacknowledged and unmanaged use of AI for translation in public services cannot continue.”
The report disclaims that not all sectors covered in the study are necessarily or directly publicly funded, clarifying that the term “public service contexts” is used in the report to emphasize the sectors’ potential to serve and affect all members of society.”
