AI in the Translation Industry – The 5-10 Year Outlook
Artificial intelligence (AI) has had a major and positive impact on a range of industries already, with the potential to give much more in the future. We sat down with Ofer Tirosh, CEO of Tomedes, to find out how the translation industry has changed as a result of advances in technology over the past 10 years and what the future might hold in store for it.
Hello Ofer, please can you tell us how technology has impacted the translation sector over the past decade?
Translation services have felt the impact of technology in various positive ways during recent years. For individual translators, the range and quality of computer-assisted translation (CAT) tools have increased massively. A CAT tool is a piece of software that supports the translation process. It helps the translator to edit and manage their translations.
CAT tools usually include translation memories, which are particularly valuable to translators. They store sentences and their translations for future use and can save a vast amount of time during the translation process. This means that translators can work more efficiently, without compromising on quality.
There are myriad other ways that technology has helped the industry. Everything from transcription to localization services has become faster and better as a result of tech advances. Even things like contract automation make a difference, as they speed up the overall time taken to set up and deliver on each client contract.
Also Read: Top 9 SaaS Startups in India 2020
Have there been any negative effects associated with the advance of technology, so far as your translation company is concerned?
Machine translation is an issue that affects not just our translation agency but the industry as a whole. Human translation still outdoes machine translation in terms of quality but the fact that websites that can translate for free are widely available has tempted many companies to try machine translation. The resulting translations are not good quality and this acceptance of below-par translations isn’t great for the industry as a whole, as it drives down standards.
There were some fears around machine translation taking over from professional translation services when machine learning was first used to move away from statistical-based machine translation. However, those fears haven’t really materialized.
Indeed, the Bureau of Labor Statistics is projecting 19% growth for the employment of interpreters and translators between 2018 and 2028, which is well above the average growth rate.
Instead, the industry has adapted to work alongside the new machine translation technology, with translators providing post-editing machine translation services, which essentially tidy up computerized attempts at translation and turn them into high-quality documents that accurately reflect the original content.
You mentioned a move away from statistical machine translation. How are machines now approaching language learning?
It was the introduction of neural networks that really took machine language learning to the next level. Previously, computers relied on the analysis of phrases (and before that, words) from existing human translations in order to produce a translation. The results were far from ideal.
Neural networks have provided a different way forward. A machine learning algorithm is used so that the machine can explore the data in its own way, learning and progressing in ways that were not previously possible. What is particularly exciting about this approach is the adaptability of the model that the machine creates. It’s not a static process but one that can flex and change over time and based on new data.
Also Read: Vectors of Innovation with Conversational AI
What does the future hold for the translation industry in terms of the impact of technology?
I think the fears of machines taking over from human translation professionals have been put to bed for now. Yes, machines can translate better than they used to, but they still can’t translate as well as humans can.
I think that we’ll see a continuation of the trend towards more audio and video translation.
Video, in particular, has become such an important marketing and social connection tool that demand video translation is likely to boom in the years ahead, just as it has for the past few years.
I’ve not had access yet to any Predictive Intelligence data for the translation industry, unfortunately, but we’re definitely likely to experience an increase in demand for more blended human and machine translation models over the coming years. There’s an increasing need to translate faster without a drop in quality, for example in relation to the spread of coronavirus. We need to ensure a smooth, rapid flow of accurate information from country to country in order to tackle the situation as a global issue and not a series of local ones. That’s where both machines and humans can support the delivery of high quality, fast translation services, by working together to achieve maximum efficiency.
AI has had a major impact on the translation industry over the past ten years and I expect the pace of change over the next ten to be even greater, as the technology continues to advance.