Will human simultaneous interpreting withstand AI for long?

Artificial intelligence has revolutionised spoken translation in recent years. Tools such as MLTM™ 4.0 and solutions integrated into videoconferencing platforms can now translate conversation in real time in over 150 languages, often with a turnaround time of less than a second. This speed and scalability are unprecedented: multilingual meetings can now be held without the systematic use of human interpreters, with reduced costs and increased accessibility.

Example #1:
At an international conference, an AI system can subtitle speakers’ speeches live in several languages and even render them orally via a computer-generated voice. This technology facilitates communication between participants from all over the world, without requiring the physical presence of an interpreter

Example #2:
In international companies, AI makes it possible to automate the translation of phone calls or discussions on collaborative platforms, making day-to-day communication smoother and more efficient.

Despite these advances, AI still faces several obstacles when it comes to simultaneous interpretation. The first challenge lies in performing in real time: in critical contexts (courts, hospitals, diplomatic negotiations), a delay of a few seconds or an interpretation error can have dramatic consequences. Yet even the most powerful AI still generates unacceptable latencies and misunderstandings in these situations

Example #3:
During a court hearing, a witness uses an idiomatic expression or cultural metaphor. An AI system could translate literally, causing an embarrassing misunderstanding or even a miscarriage of justice. A human interpreter, trained to understand implicit meaning, will adapt his or her translation to faithfully reproduce the meaning and intention of the witness.

Example #4:
In a medical setting, a patient explains his or her symptoms using regionally specific expressions (such as “I’ve got pins and needles in my legs”). Literal interpretation by AI would risk misdiagnosis, whereas a human interpreter would be able to translate the expression to provide the appropriate medical context.

One of the major strengths of human interpreters lies in their ability to capture and convey cultural, emotional and contextual nuances. An interpreter doesn’t just translate words: he or she restores intonation, silences, hesitations, humour and even irony. These subtle elements still largely elude machines, which struggle to grasp the human dimension of communication.

Example #5:
At a press conference, a speaker makes a joke or an allusion to a cultural event. The human interpreter will adapt his or her translation so that it is understood by the target audience, whereas AI interpretation might miss this subtlety or make it incomprehensible.

Example #6:
In business negotiations, a participant uses a sarcastic tone or makes an implicit allusion. The human interpreter will understand the intention and adapt his or her speech to convey the message with the same force, whereas AI interpretation could misinterpret the nuance and create a misunderstanding.

Confidentiality is another major issue. Interpreters often work with sensitive information (industrial secrets, victims’ testimonies, medical data, etc.). Mistrust of cloud solutions or the risk of data leaks are hindering the adoption of AI in this sector.

Example #7:
During a strategic corporate meeting, the confidentiality of the information shared is paramount. Companies often prefer to use a human interpreter, with proven ethics and discretion, rather than an automated solution which could store or transmit data to third parties.

Rather than competing with each other, AI and human interpreters will be able to work together more closely. AI can assist the interpreter in terminology preparation, automatic subtitling and noise management, while humans can focus on rendering nuances and managing complex situations.

Example #8:
At a technical conference, AI can provide specialised glossaries or definitions in real time, enabling the interpreter to become more accurate and efficient. Automatic subtitling can help the hearing-impaired and provide a written record of speeches.

Example #9:
In a noisy environment, AI can filter out ambient sound to improve audio quality and facilitate the interpreter’s work.

Human expertise remains irreplaceable when it comes to ensuring the quality, safety and comprehensive nature of communication. Although AI is making great strides, it is not yet capable of reproducing the variety, versatility and subtlety of human communication. Human interpreters will therefore continue to play a key role in demanding fields, in which an in-depth understanding of context and creative adaptation are essential.

Example #10:
At a diplomatic summit, the slightest misinterpretation can have international consequences. Heads of state prefer to rely on human interpreters, with their proven ability to respond to and understand the issues at stake, rather than on an AI which could misinterpret a nuanced remark or a political reference.

A future of sustainable co-existence

In conclusion, human simultaneous interpreting will withstand AI for a long time to come, as technology is not yet able to reproduce the variety, versatility and subtlety of human communication. The future of the profession lies in human-machine complementarity: AI can assist interpreters to improve the quality and speed of their service, while humans ensure comprehensive, confidential and creative communication.

At Birdwell, we firmly believe in the added value of human expertise, while integrating the most powerful AI tools to offer our customers cutting-edge translation and interpreting solutions. Contact us to find out more about our tailor-made services and make the most of both worlds: the speed of technology and the sophisticated skills of humans.