AI and Translation: Reflections from AI Voices II, CIOL Translators Day, and the Future of Real-Time Interpretation

by Vasiliki Prestidge
 

AI continues to reshape expectations and norms in the translation and interpreting industry at a staggering pace, and resources such as the CIOL AI Voices II publication have captured some of this momentum and the anxiety and concerns that the pace of change can trigger. Released as a follow-up to the original CIOL AI Voices in 2023, this second edition digs deeper into the evolving relationship between language professionals and AI technologies based on two years of developing practice and evolution in the capabilities of Generative AI. 

AI both challenges and enhances the work of linguists

What stands out for me in AI Voices II is its balance. Rather than presenting AI as an existential threat or a ‘silver bullet’, it explores the nuanced ways in which AI can both challenge and enhance the work of linguists. Contributors offer real-world experiences and critical reflections, including the emerging practical and ethical complexities we increasingly navigate in terms of client expectations and our own ethics and values. It’s a timely and important read for anyone working in language services today.

AI in the room – a conversation

I also spoke this year at the Chartered Institute of Linguists Conference Season. CIOL Translators Day, with over 200 attendees, provided a vibrant platform for continuing this conversation and the opportunity to network and hear how others are experiencing the impacts of AI. In every session it felt like AI was in the room – even if that wasn’t the subject – underscoring just how central this topic has become to the profession. What struck me most was the diversity of perspectives represented, from those who are embracing AI as a productivity tool to those urging caution over data privacy and quality assurance and even an ethical and environmental imperative not to use AI at all. The sessions focused on uniquely human skills – cultural, translating archives and building customer relationships, were met with a sigh of relief by many. AI can cast a long shadow.

Effects of AI on other sectors

For many translators, AI is no longer a far-off innovation – it’s an everyday companion, whether they like it or not. The CIOL Council ‘Expert panel’ brought a wider perspective, referring also to the impacts of AI in other sectors, other developments affecting the industry, including changes in education and the geopolitical context, flagging that AI may not be the only cause of an industry and a profession facing up to change.
 

AI in interpreting

As for recent developments, one that has particularly caught my eye is Google's unveiling of real-time live interpreting capabilities in Google Meet, powered by its Gemini AI. Demonstrated at the Google I/O 2025 conference, it was presented as follows: 

“This feature allows spoken words in one language to be simultaneously interpreted into the conversation partner's preferred language, preserving the speaker's original voice, tone, and inflection.”

While the innovation is impressive, it's important to note that the demonstration focused on interpreting between English and Spanish – languages that share many structural similarities and which are commonly translated and heavily present in AI training data.

For me there remain unanswered questions about the system's performance in more complex scenarios. For instance, how would it handle high-stakes situations such as conference or diplomatic meeting involving languages that are structurally and culturally distant, with less parallel data – say interpreting between Greek and Chinese. The challenges in such cases are multifaceted, involving not just linguistic differences but also cultural nuance and terminology that require precise interpretation and real-time judgement in how best to render them in another language.
 

The importance of human intervention

Moreover, as Google acknowledges, while the technology shows promise, it's not without its limitations. Early tests have revealed issues such as lag, misinterpretations, and awkward phrasing – especially at the beginning of sentences. The system also begins translating before a speaker finishes, which presents challenges with context and word prediction especially in grammatically different languages. These shortcomings really underscore for me the importance of human oversight, particularly in higher stakes scenarios where accuracy is paramount.

Conclusion

To conclude, AI Voices II and the CIOL Conference have made one thing clear: AI is no longer on the periphery of translation, it’s at the centre. And, with innovation in spoken word tools, it is likely to become more present in the world of interpreting. But using AI doesn’t have to mean replacement. With thoughtful adoption and ongoing client dialogue, I firmly believe we can shape AI as a tool that enhances, rather than undermines, the human expertise at the heart of our profession.

The CIOL AI Voices II is available on the CIOL website here: CIOL AI Voices II

Vasiliki Prestidge BA MA MCIL CL is a member of CIOL Council, a translator, interpreter, transcreator, blogger, consultant and director of Greek to Me Translations Ltd. Read her full CIOL Council biography here.  

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