As in many other sectors, AI (artificial intelligence) has developed strongly in the translation sector in recent years. So what is the impact of AI on the translation profession? What are the advantages and limitations of this technology? Alphatrad has the answers to these questions.
AI in translation: definition
AI first appeared in the 1950s. It refers to the methods and tools used by a machine to imitate human intelligence. Using massive amounts of data this technology creates algorithms allowing it to reproduce behaviours similar to those generated by the human brain, such as reasoning, planning, creativity, etc. AI is now widely used in many fields of activity (transcription, translation, retail, banking, production, etc.)
AI can thus make use of technologies such as deep learning, big data, cloud computing or machine learning.
What are the methods of translation with AI?
Machine translation
Translation professionals have been using digital solutions such as computer-assisted translation (CAT) for many years in their profession.
Machine translation came into being about fifteen years ago, with the advent of Google Translation, which marked the beginnings of translation with AI. However, it soon became clear that this method had considerable limitations: the translation was literal and often contained gross language errors.
A distinction needs to be made between RBMD machine translation (based on linguistic rules, this method uses dictionaries to translate content) and statistical machine translation. The latter uses machine learning to operate, with algorithms capable of analysing large volumes of existing translations and extracting statistical patterns.
Neural machine translation
AI really came into its own in the field of translation in 2017 with DeepL. This neural translation software is based on data from the translation website Linguee, a multilingual dictionary that compares translations in nearly 20 languages.
This marked the beginning of neural machine translation (NMT). This intelligent technology, which is based on artificial neurons, considers the entire text and its context. Furthermore, NMT is able to continuously improve and refine itself, thanks to the flow of data it receives.
Neural machine translation is a major technological advance that allows us to obtain a text that is much more qualitative and fluid than with machine translation.
Translation software that uses AI
In addition to Google Translate and DeepL, here are some other examples of AI-powered translation software:
- Microsoft Translator: this software uses machine learning techniques and can support over 60 languages
- Systran: This AI-enabled translation software covers more than 130 languages. It uses deep learning methods to optimise the quality of translated texts
- Meta's AI translation software: Meta, the parent company of Facebook, created a brand- new AI model in 2022 that is capable of translating, in real time, more than 200 languages worldwide, simultaneously. Based on machine learning, this technology has been made freely available and is intended to cover all languages, including minority languages that are not currently covered by existing machine translation systems. It is the NLLB-200 (No Language Left Behind) model
- ChatGPT from OpenAI: Although not a full-fledged translation software, the open source ChatGPT software is capable of translating words, sentences, or entire texts in many languages, thanks to high-performance AI techniques.
What are the advantages and limitations of AI in translation?
Here is an overview of the main advantages and disadvantages of translation with AI:
The benefits of AI in translation
AI allows for extremely fast, high-volume translations that are increasingly accurate and precise. AI-enabled translation software that uses machine learning has the ability to self-correct and improve the quality of the translations generated.
In addition, many AI-enabled translation tools are capable of translating multiple texts in different languages simultaneously.
Finally, these methods have the advantage, for users, of being very low cost (or even free) and of covering a very wide range of languages.
The drawbacks of AI in translation
Despite its many advantages, AI is far from infallible and still has many limitations in the field of translation. Indeed, this technology is not able to adapt the translation to the target readership.
Nor can it consider local cultural norms and customs, the clients’ expectations, the style, the translation's intention... These are essential elements in translation, to obtain texts that are respectful of the local culture, adapted to the target audience and faithful to the source text.
Furthermore, while AI translation can be effective for the most common languages (English, French, Spanish, German, Dutch, Italian, Arabic, etc.), it is much less effective for rare languages or dialects for which little data exists. In such cases, the AI will very often have to use English translation as an intermediate step, which may generate significant errors and misunderstandings.
What is the impact of AI on the translation profession?
Given its many limitations, it is unlikely that AI will completely replace human translators. Indeed, in the case of technical, marketing, financial, legal, or medical content, the slightest error or ambiguity can have damaging consequences, generate disputes, and have a major impact on the company's reputation.
The expertise of a native translator, their knowledge of the local culture, customs, and cultural subtleties remain indispensable.
However, AI will undoubtedly facilitate the work of the translator. We are already seeing the emergence of a hybrid method: neural translation with post-editing, where AI is combined with human know-how. Far from being a threat, AI could therefore end up being a real opportunity for translators.
and receive an answer within a few hours
- Choose the service you require
- Get a quote
- Confirm and receive your order
Our answers to your questions about translation and AI
The exact definition of AI given by the European Parliament is as follows: “AI is the ability of a machine to display human-like capabilities such as reasoning, learning, planning and creativity". AI is therefore not a technology as such but rather a scientific concept, which consists of developing technologies capable of imitating certain human abilities.
With the rise of AI in translation, its performance and very low costs, many translators are concerned that AI poses a threat to the future of their profession. But it seems highly unlikely that AI will completely replace human translators, whose expertise remains unmatched until now. To remain competitive, however, professional translators will have to learn to work with AI, including by offering post-editing services for machine translations.
Neural machine translation is the most accurate and efficient method of translation with AI. It has the ability to mimic the human brain, to question itself and to consider the context of the source text. It is also capable of evolving over time and improving autonomously.