By Dom Hebblethwaite, CIOL Head of Membership

The role of generative AI (ChatGPT and other LLMs) in professional translation and language services remains controversial, especially when it is used alongside Machine Translation in ways that potentially impact translation quality, linguists' earnings, and professional satisfaction.

However, what about when technology is not imposed by clients but instead adopted voluntarily as a personal productivity tool by linguists themselves?

CIOL's recent survey, with 134 respondents, reveals that AI divides practice as well as opinion. While some language professionals are embracing AI-powered tools, others remain sceptical or outright reject their use. The results highlight the varying attitudes toward and experiences of generative AI within the profession, as is increasingly the case for many 'knowledge workers' in all sectors.

The Poll Results at a Glance

The survey results show an almost 50/50 split between those who use AI and those who do not:
 

• 49% of respondents use generative AI either professionally (31%) or for personal tasks (18%)
• 51% of respondents do not use generative AI - either having tried but discontinued use (13%) or never having used it at all (38%)

These figures suggest that while many are using generative AI in the profession, a substantial proportion of translators and linguists remain hesitant, if not outright opposed, to its adoption and use.

AI as a Research Tool

One respondent noted that they "use AI predominantly for research", indicating that generative AI's role in supporting translation extends beyond direct text generation or checking.

Another linguist specifically mentioned: "I would only use AI really for background information and queries rather than the actual translation".

Some take advantage of built-in AI tools, with one respondent sharing: "I use Google Chrome-generated AI when researching terms/phrases".

These comments suggest that AI-powered tools can assist with background research, terminology verification, and information retrieval, helping translators streamline their workflows. This aligns with broader trends in the industry, where AI is increasingly valued as an auxiliary 'helper' tool rather than any thought of it as a direct replacement for human expertise.

The Practicality of AI in Translation

Some respondents have experimented with AI tools but found them impractical for their specific needs. One individual mentioned trying CoTranslatorAI, a tool designed to interact with OpenAI for post-editing machine translation (PEMT). However, they found it cumbersome to set up and less effective than DeepL for generating rough drafts.

They also voiced concerns about the attitudes of those developing such tools, stating: "The people that run it seem to think they're techbros with Musk-like aspirations, and I don't want to encourage them".

This reflects a broader frustration and concern amongst language professionals, many of whom find new AI tools overhyped and impractical, particularly when existing solutions like DeepL already provide effective machine translation support. The underlying concern here is not just about functionality but also about the culture and ethos driving AI development in the translation and wider languages space.

Resistance to AI: Personal and Ethical Considerations

For some linguists, generative AI simply holds no appeal.

One respondent candidly admitted: "I am not really interested. Too close to the end of my career, I am probably over-complacent, but hope I can get by with a CAT tool and the fact that I am human as my selling point. I am told AI uses enormous amounts of energy and is not green, and hope to get by without it!"

This response highlights two themes:

1. The hope that human expertise and traditional CAT (Computer-Assisted Translation) tools will remain sufficient for fulfilling and high-quality professional translation work
2. Concerns about the policies, practices, and attitudes which drive much AI development and the environmental costs related to the high energy consumption of AI models and social costs related to their large-scale deployment

The environmental and social impacts of AI are particularly significant for some translators. One respondent explicitly stated they "Do not use on account of concerns around confidentiality, ethics, environment and questionable usefulness".

Another shared a similar perspective: "I read about the eco-damaging impact of using ChatGPT so decided to reduce use and then ended up using it so little that I didn't feel I needed it right now".

Such concerns are valid; some linguists who are well established in their careers may not see the value in investing new time and effort into learning AI-based tools that they might not need. Moreover, the social and environmental impacts of AI remain an issue for us all.

The Generational Divide and the Loss of Traditional Skills

Another respondent expressed deep concern about the impact of AI on the translation profession: "Machine translation gives the illusion of coherence and clarity, and post-editing it requires very different skills to those I have honed over 30 years' practice. I suspect this is the position many linguists of my generation find themselves in and that many of my cohort, like myself, are now semi-retired. I am very worried about the potential loss of basic translation skills and experience being handed down to and shared with the next generation".

This sentiment is echoed by many experienced translators who fear that AI-driven translation workflows may devalue the core skills of translation, reducing the need for linguistic expertise in favour of post-editing machine-generated text. The shift from crafting nuanced translations to simply refining AI output represents a fundamental change in the profession—one that some feel threatens the depth of linguistic skill and undervalues the knowledge traditionally passed down through mentorship and training.

What Does the Future Hold?

The results of this poll highlight a potential bifurcation in the translation profession. Generative AI is here and, across all knowledge-based professions, its role is growing, but its adoption is far from universal. While some professionals see AI as a valuable tool for research and productivity enhancement, others remain either disinterested or actively resistant, given their concerns over usefulness, ethics, and the preservation of linguistic skills.

For the profession as a whole, the key challenge in the medium-term likely lies in striking an acceptable balance. Can AI be integrated in a way that supports productivity and enhances human expertise rather than diminishing it? Can language professionals find sustainable ways to work with AI so that it complements and enhances their skills rather than replacing them? And importantly, how can new generations of 'digital native' linguists who have grown up with AI be mentored, trained, and supported to preserve the core competencies and values of traditional translation and language work?

Conclusion

The CIOL poll underscores a reality that is unlikely to change overnight: generative AI is and is likely to remain an ever more contentious issue in the translation industry. For some, it represents a valuable new set of tools in the toolkit that can enhance productivity and streamline workflows. For others, it is an impractical, ethically questionable, and potentially damaging force that threatens the very fabric of the profession.

For all the challenges AI poses, complete resistance may not be the most productive or realistic stance. Linguists likely need to be 'AI-curious' — both to potentially benefit from AI as a productivity tool and also to be able to critically evaluate and highlight to clients and stakeholders its shortcomings. By understanding what AI can and cannot do well, language professionals can better position themselves as the go-to experts in what continues to be a rapidly evolving field with as much hype as reality.
 

Dom Hebblethwaite is the 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.