Expert Perspectives: What AI Means for Language Quality Management [Video]

How can we harness the power of AI to improve linguistic accuracy, increase efficiency, and more? To explore this question, Beyont held a panel discussion with two guest experts on today’s technological wave. 

We invited our panelists to shed light on the evolution of language quality management and how AI is transforming the field. Read on for essential takeaways, and watch the full video for more in-depth insights. 

Meet the Experts

Konstantin Dranch is a language industry researcher and co-founder of Custom.MT, an AI and machine learning agency working with localization and translation teams. 

Stefan Huyghe is VP of localization at Communicaid Language Solutions and the creator of AI in LOC, a newsletter about AI and the language industry. 

Cesar Matas Alsina, language quality program manager at Beyont, led an in-depth talk followed by audience Q&A. 

Key Takeaways: What AI Means for Language Quality Management

Many ideas and insights emerged from this far-ranging conversation. Let’s look at just a few. 

1. Automation is becoming a reality. 

Language quality metrics are subjective and vary with each organization’s needs. As a result, linguistic QA has been a difficult process to automate. But that’s changing with AI. Quality prediction models are growing in power, and translation management systems are integrating AI-powered QA tools. Teams can now press a “magic button” to rate the quality of translations (and possibly even resolve disagreements among linguists). 

2. Democratization of AI will rewrite the rules of the game. 

For example, OpenAI’s new custom GPTs allow users to create their own version of ChatGPT and customize it to their needs. That means anyone can develop a system for automated linguistic QA—for example, for translations from English to French—in a few simple steps. Powered by the latest models, such tools could help language quality teams boost efficiency while bypassing years of development. 

3. AI can simplify fine-grained evaluations of language quality.

Unlike machines, humans can only remember and manage a handful of variables at a time. Language quality measurement tools such as MQM-DQF, with its complex system of error categories, are thus hard to learn and require extensive training. An AI can apply metrics with many more parameters, helping linguistic QA teams categorize errors more accurately and consistently. 

4. Accuracy remains a significant challenge. 

AI language models can only learn properly by training on large amounts of high-quality data. However, this kind of data is hard to find for many languages other than English. Current AI can do a job similar to a competent student, but it’s not yet on par with expert linguists and editors. For the foreseeable future, a human evaluator will still need to double-check the results. 

5. Linguists will take on new roles in an AI-powered world. 

As the localization industry shifts toward data-driven products and services, linguistic experts may move into unexpected roles, such as training new AI engines. AI will make it more cost-effective for companies to create content in less widely translated languages—so there may be increasing demand for specialists who can perform annotation and fill gaps in the linguistic data. 

AI’s Role in Language Quality: Catch the Complete Discussion

If you work in language quality or localization, you need to know where AI is taking the industry. Want a leg up? Watch the full-length video to get more expert insights and see into the future of language quality management. 

You can also watch this video and all past videos as well as subscribe to our Youtube channel using the button below!