Chartered Institute
of Linguists
Responsible use of AI in translation: why critical engagement matters
By Mariam Aboelezz, CIOL Vice Chair
Artificial intelligence is not an abstract or distant prospect for translators and interpreters. It is already embedded in the tools we use, the workflows we are asked to adopt, and the expectations many clients now bring to the table. Machine translation is here, it will remain here, and it will continue to evolve. The real question is not whether we use AI, but how we use it – and whether we do so responsibly.
I was reminded of this very clearly when I recently introduced a postgraduate module on specialised translation and technical skills. I wanted to make space for students to engage with AI openly, ethically and, above all, critically. Avoidance is not an option. Nor is unthinking adoption. As linguists, we need to develop informed, reflective practices that allow us to harness technology without surrendering professional judgement or ethical responsibility.
Transparency is non‑negotiable
The starting point, for me, is transparency. If AI or machine translation is used at any stage of the translation process, that use should never be hidden or misrepresented. This applies just as much in professional practice as it does in academic settings. Passing off machine‑generated output as wholly human work undermines trust – with clients, with institutions and with our profession as a whole.
There are contexts where machine translation can be genuinely useful. The key is ensuring that clients are part of that decision, that expectations are clearly managed, and that responsibility for the final product remains firmly human.
Environmental impact: an ethical blind spot?
One area that is still too often overlooked is environmental impact. Generative AI systems are resource‑intensive, with significant demands on energy and water. This does not mean they should never be used – but it does mean that we should use them thoughtfully, efficiently and with awareness of their wider consequences.
Learning to work well with AI, rather than repeatedly prompting and re‑prompting, is not just a matter of productivity. It is also a matter of sustainability. As linguists, we cannot afford to ignore the environmental cost of the technologies we adopt.
Copyright, ownership and professional risk
Copyright is another area where assumptions can be dangerous. In the UK and many other jurisdictions, original human translations are protected by copyright. Machine‑generated translations are not. Hybrid outputs may also fall outside legal protection, leaving translators exposed.
Equally important is the question of what happens when copyrighted material is fed into AI systems. Without a clear understanding of copyright law, translators may inadvertently place themselves – and their clients – at risk. Staying informed is no longer optional; it is part of professional competence.
Privacy, confidentiality and sector‑specific realities
For linguists working in legal, medical or other sensitive domains, privacy and confidentiality must always take precedence. Patient data, legal documents and confidential materials should never be entered into machine translation systems unless robust safeguards are in place – and often not even then.
The fact that some clients explicitly prohibit the use of machine translation in certain contexts is not technophobia; it is a recognition of legal and ethical realities. Respecting those boundaries is fundamental to professional integrity.
Bias, prompts and the myth of neutrality
AI systems are not neutral. They reflect the data they are trained on and the choices made by their designers. Some platforms may perform better environmentally while performing worse in terms of bias. Others may produce fluent output that conceals problematic assumptions.
This is where prompt engineering becomes more than a technical skill. Knowing how to ask the right questions – succinctly, precisely and critically – can help reduce bias, improve output quality and limit unnecessary resource use. More importantly, it keeps the human firmly in the loop as an active decision‑maker rather than a passive recipient.
Regulation, guidelines and professional responsibility
Regulation is evolving rapidly, and unevenly, across jurisdictions. Copyright law, data protection and AI governance are all areas where “playing catch‑up” is the norm. This places a responsibility on individual practitioners to stay informed, particularly when working across borders.
Professionals have a crucial role to play here, and I would strongly encourage linguists to engage with sector‑specific guidelines and resources. AI is not a challenge unique to our profession, but it is one that we must address on our own terms.
Staying professional in a moving landscape
I deliberately think about these issues as headlines or guidelines rather than fixed rules. The technology is moving too fast for static answers. What matters is cultivating habits of critical engagement, ethical reflection and lifelong learning.
AI will change how we work. It does not have to change who we are as professionals.
Dr Mariam Aboelezz is a Vice Chair of CIOL Council and a Lecturer in Arabic Translation Studies at the University of Liverpool. Prior to joining the University of Liverpool, she taught at University College London, Birkbeck College and Lancaster University.
She is also a practising translator and most recently worked as an Arabic translator at the British Library for seven years.
Mariam holds a PhD in Linguistics and MA in Language Studies from Lancaster University, and a BA in English Language from Ain Shams University in Egypt. She is also a Chartered Linguist and holds the CIOL Qualifications Diploma in Translation (DipTrans).
Views expressed on CIOL Voices are those of the writer and may not represent those of the wider membership or CIOL.
Filter by category
More
