From courtrooms to hospitals, interpreting demands more than language fluency – yet experts warn AI is changing how the profession is understood and valued.
When more than 200 interpretation errors emerged in a Victorian Supreme Court trial, the issue was not simply technical. Lawyers argued the mistakes distorted evidence given by an Arabic-speaking witness, raising concerns about fairness in the judicial process. Proceedings continued only after transcripts were reworked with a second interpreter's review.
For those working in translation and interpreting, the case is a stark reminder of what is at stake.
While the Victorian case did not involve AI, experts say it highlights the complexity and responsibility embedded in interpreting work at a time when artificial intelligence is reshaping the industry.
They warn language translation is not a simple automation task where word replaces word – and that misunderstanding the profession risks real-world consequences.
"People keep asking what the future holds," says Professor Ludmila Stern, an expert in translation and interpreting from ADA's School of Humanities & Languages.
"There is a lot of concern about AI, but I don't believe the profession will disappear. What we need to be careful about is how these technologies are used."
Prof. Stern is the founder of UNSW's Master of Interpreting and Translation, now in its 20th year, and has played a key role in shaping translation and interpreting education in Australia.
Machine translation tools and large language models are now embedded across the industry, particularly in written translation.
But UNSW Professor Sandra Hale, a pioneer in legal interpreting research and one of the main contributing authors of the National Standards for Working with Interpreters in Courts and Tribunals, says this growing reliance reflects a misconception of the profession itself.
"A lot of people assume it's just about knowing another language," she says. "But interpreting involves analytical, cultural and ethical judgement. You're making decisions constantly."
Unlike written translation, interpreting is done in real time, and interpreters must process and deliver meaning instantly while managing nuance, tone and context. Those demands are especially difficult to achieve in high-pressure environments such as courtrooms, hospitals and police interviews.
"You're not just transferring words," Prof. Hale says. "You're interpreting meaning – not just what is said but also how it is said, to achieve the same effect in the listener as the original message – and you have to do that instantly."
At the same time, demand in interpreting is rising. Australia's linguistic diversity continues to expand, placing pressure on systems already stretched. Courts alone may require interpreters across hundreds of languages, from migrant communities to Indigenous languages.
"The demand keeps growing," Prof. Stern says. "That creates real pressure on the system."
Yet the workforce has struggled to keep pace. Training pathways and working conditions remain uneven, and qualified practitioners are not always available when needed.
"We know there are situations where untrained people are used, or where conditions are less than ideal," Prof. Hale says. "That can affect quality, even for very experienced practitioners."
For Prof. Hale, the risks are clear. "Court interpreting is a very high-stakes job. It requires highly trained professionals and proper conditions – otherwise errors will occur, with serious consequences."
Where AI helps and where it falls short
Technology is both part of the solution and part of the challenge. For experienced translators, AI tools can improve efficiency and help manage large volumes of text. But Prof. Hale emphasises that these systems cannot replace professional judgment.
"A machine doesn't understand meaning, it processes patterns," she says. "So whatever it produces still needs to be checked by an expert translator."
Without that expertise, errors can easily slip through. Studies have shown that people without formal training often struggle to detect inaccuracies in machine-generated translations.
"There's a real danger in assuming the technology is always right," Prof. Hale says. "If you're not trained, you may not even realise when something is wrong."
Beyond accuracy, AI also raises ethical concerns. Because systems are trained on existing data, they can reproduce biases embedded in language over time.
"We've made progress in addressing bias in language, but AI can reintroduce it," Prof. Hale says. "It reflects what it's been trained on."
There are also risks around confidentiality. Many free online tools store or process data externally, making them unsuitable for sensitive material.
"You can't just upload confidential documents into free systems," Prof. Hale adds. "That's a serious breach of the professional ethical requirement of confidentiality."
Integrating AI into translation and interpreting practice
For Prof. Stern, the question is no longer whether AI will be used, but how. In areas such as media subtitling and publishing, AI-assisted workflows are becoming standard, but always with human oversight.
"The important thing is that there is always a human in charge of the final product," she says. "That's what guarantees accuracy and accountability."
Interpreting, however, remains far less vulnerable to automation. Its reliance on real-time interaction, judgement and human presence makes it difficult to replicate.
"In situations where accuracy really matters, a human interpreter is still essential," Prof. Stern says. "At this stage, there is no real substitute."
The pandemic accelerated other shifts. Remote interpreting, particularly via video, is now commonplace. While it has improved accessibility, it has also introduced new challenges – especially when services rely heavily on audio alone.
"Telephone interpreting is often seen as a cheaper option, but it comes at a cost," Prof. Hale says. "Without visual cues, communication becomes much harder, and the quality can suffer."
As the profession evolves, education remains central to ensuring high standards. Training programs have adapted to incorporate new technologies, expand language offerings and respond to emerging industry needs.
"Our [UNSW] Program has always been closely linked to research and practice," Prof. Hale says. "What we learn feeds directly into how we train students, and that in turn supports the profession. There is a cross-fertilization between research, training and practice."
Stern says that adaptability has been essential as the linguistic landscape shifts.
"When we started two decades ago, the needs were different," she says. "Now we're seeing a much wider range of languages and contexts. Training has had to evolve alongside that."
Despite the pace of change, both scholars remain optimistic about the future. While AI will continue to reshape workflows, they argue the profession's core purpose—enabling meaningful communication across languages—remains unchanged.
"At the end of the day, it's about enabling people to understand each other," Prof. Stern says. "That's something that still relies on human judgement, empathy and responsibility."
For Prof. Hale, that human element is precisely what ensures the lasting relevance of the field.
"It's an incredibly demanding profession," she says. "But that's also what makes it essential, especially in a world where clear and accurate communication matters more than ever."
