This past year saw the rise in popularity of a new form of Artificial Intelligence (AI) that was made available to the public, often free of charge. Open AI’s ChatGPT, Google’s Gemini and others have brought the power of robust computer models trained on more data than which the public has access to the average person, giving them access to power information and tools. Recently the Chartered Institute of Linguists (CIOL) conducted a survey among CIOL members about changes in work in the translation and interpreting industries in 2024 related to AI and Machine Translation (MT) use.

Survey Results

The diagram above represents the changes in work volume that CIOL members who took the survey reported in 2024. More than half reported some reduction in work volumes (number of projects and/or total number of words to translate per project), with nearly 50% of all repsondents showing a significant decrease in work volume. Among all groups, including those that reported increases in work volumes, the total percentage of Machine Translation Post-Editing (MTPE) – where a tool like Google Translate first translates a document and then a human edits that output – increased dramatically in 2024.

Implications and Changes in Work

While the diagram shows stark results as a freelancer, the implications for the field is terrifying. Several respondants reported stopping freelance activities in the past year; working as a freelancer has become unsustainable when there is little work. MTPE is usually paid at a much lower rate than traditional translation.

It is also not just translators that are feeling the pinch. Translation agencies, especially those that are committed to high-quality, human translations, have also suffered as end clients want translations faster than before. To cope, they use AI or MT without realising (or being able to realise if they do not understand the arget language) the drop in quality compared to human translations. Some translations agencies have switched entirely to MTPE; many more have simply gone out of business.

For those still in translation, there may however be a silver lining. More translators reported working with direct clients (i.e. cutting out the translation agency “middle man”) specifically offering high-quality, human translations; some have reported being able to supplement their income as they can charge more than they would receive from an agency without having to pay the same upkeep and maintenance costs that a traditional agency has.

My thoughts

In the groups above, my data is unfortunately included in the “signficant decrease” group. In 2024 I saw the volume of work drop by at least half and I also outlived two different translation agencies that I worked with – one went completely out of busines and the other merged with a larger company. Nowadays the vast majority of my translation projects have been MTPE, so I have suffered a double whammy of less work overall and less renumeration per project. On top of this, the price per word (i.e. typically standard payment metric in the translation word, although some language combinations use different metrics such as “price per line” – a line = 55 characters including spaces – or “price per character” for many Asian scripts that do not have spaces) for human translations have also decreased as many translation agencies can no longer pay higher prices due to their own reduction in income.

My hope is that in the near future, people are going to realise the privacy concerns of using AI and MT with translation. As I wrote in a previous article on Machine Translation Use in Public Services: A Roundtable Discussion (unrelated to this survey but also heavily focused on changes in work in public services in the UK and other countries), AI and MT store the data that is input into it. For things like greeting cards, there are no problems using AI/MT, but for medical, legal, scientific and business fields among many others where private information cannot be shared with the public, using AI/MT connected to a public database is a data breach. However, many people do not seem to understand or connect this fact with translations.