In our latest poll, members and fellow linguists have told us how they use generative AI models, the results are above. And here is how that picture compares to last year's poll:

The headline number is that 'users' now outnumber 'non-users' for the first time, and the growth has come from the 'professional use' end of the spectrum. The 'personal only' and 'tried and stopped' groups have both shrunk a little. The 'never used' group has come down too, but only by five points.

Small changes, big difference

If the 2025 poll suggested an evenly divided profession, this year's comments suggest something more nuanced.

Users are getting more specific about what AI is actually good for. Non-users are getting more articulate about why they want no part of it. The middle ground, the "I tried AI, it was fine, I'm not sure" position, is the bit that's thinning out.

What linguists are actually using AI for

The biggest cluster, by some distance, is research and terminology. Very few linguists describe using AI to produce translations.

Many describe using it as a kind of fast but fallible reference tool:

"I use ChatGPT and Perplexity for terminology questions, and Claude for summaries of large documents and for writing tips and improvements. I do not use it for actual translation."

"I never use generative AI directly to produce a translation. However, I do use Perplexity like a thesaurus, asking it to look for published examples of different phrasings, and also sometimes for research. I choose Perplexity because I've heard (rightly or wrongly!) that it's better at displaying its sources and less likely to hallucinate."

"I use AI to discuss terms and phrases in the source text, and to find synonyms in the target text. I don't use AI to do the actual translation."

A second cluster is brainstorming and rephrasing, particularly for creative or marketing work:

"Brainstorming, rewording, finding ideas for alternatives or puns in marketing texts."

"I sometimes use generative AI for marketing, content writing, SEO writing, and research, and mostly for personal branding."

Interpreters describe a specific use case in preparation:

"I use ChatGPT in preparation for interpreting, for example to generate lists of terms and proposed translations relevant to upcoming interpreting assignments. On rare translation assignments out of my native language, I use ChatGPT to give me ideas for translating terms and phrases. It is far better than Google Translate for context. The final decision is mine!"

Teachers describe using it for lesson planning and examples:

"I use Gen AI to help with lesson planning, especially to create examples of specific teaching topics. Also in editing, to check correct punctuation and usage."

What runs through almost every 'user' comment is a careful caveat about AI’s role. One respondent put it as cleanly as anyone:

"I use Gen AI as a guide, but the final decision regarding the best translation always rests with me as the linguist."

Why some have stepped back

The 'tried but discontinued' group is small (9%) but the comments are striking. Disappointment with the output features, but ethical and environmental reasoning often comes through more strongly:

"I used them mainly for brainstorming, especially for the translation of culture specific or complex contexts. But the ethical burden (how AI is created using stolen art and human labour, the immense amount of resources it uses, the affiliations and projects of some companies, etc.) became too much for me, so I stopped using them altogether."

"I tried using them because there were so many articles and talks saying how amazing and useful they were. I was very disappointed and found that they were not useful at all. Since then, I learned about the environmental and social costs of Gen AI, so I decided never to use them again for ethical reasons."

One technically-minded member who had used a specific PEMT tool put their reason for stopping more bluntly:

"Dabbled with CoTranslatorAI a couple of years ago, but stopped using it as the guys that run it are an ethics-free zone. Later discovered of course that AI in general is unethical, based on theft, and destructive to the environment, so I also stopped dabbling with it outside of work too."

The 'never used' third

A third of respondents have never used generative AI at all.

A year ago, many 'never used' comments came across as "not for me, not interested".  This year, more of them read as an active, considered and strongly held position:

"I am against the use of AI in general and believe it to be ethically and environmentally dubious. I explicitly tell my clients that I do not use AI."

"Plenty of reasons NOT to use it: it's more work to fix the results than to translate from scratch; AI is built on stolen and copyrighted work; AI uses far too much water and electricity; AI companies use low-paid workers to edit the input; using AI decreases your creativity and skills."

"I don't see any advantage for me in my work anyway, but I'm increasingly worried about the environmental impact and serious ethical issues of AI."

And then there's the comment that captures a particular kind of anxiety that has grown over the year:

"I haven't used them, and have no desire to, but the company I work for is moving to an AI-first approach. I feel it disregards the skills and expertise of the human linguists. I am worried my job will disappear. The LLMs are getting an awful lot of free training data from people, both in work use and casual use."

Where do we go from here?

A year ago, the prevailing suggestion was that linguists – like many other knowledge workers - might benefit from being 'AI-curious': engaged enough to know what these tools can and can't do, without surrendering professional judgement to them.

The 2026 results suggest a hardening of differences in view.

For AI users: the focus is now more specific:

"where exactly does this help me, and where does it hinder?"

Many comments in this year's poll come from linguists who answer that question quite precisely: terminology yes, drafting no; brainstorming yes, finished prose no.

For non-AI users: the sentiment is:

"I want to keep working on my terms, despite a market that's increasingly going to assume AI use."

But that's built on the belief that what hasn't changed is more important than what has: the value of a skilled, accountable linguist whose judgement, ethics, and craft cannot be automated. A market saturated with AI output doesn't diminish that value; if anything, it makes it more important.

The AI divide is real, but, it has moved more from 'curiosity' and 'experimentation' to 'practicality' and 'principle';  and what’s clear from our poll is that everyone agrees that the highly-skilled, accountable linguist should remain where they have always been: at the heart of professional languages work.

Dom Hebblethwaite is Head of Membership for the Chartered Institute of Linguists. For more on Dom see his profile here. 

Views expressed on CIOL Voices are those of the writer and may not represent those of the wider membership or CIOL.