How Many Language Service Providers Use Large Language Models in Workflows?

How Many Language Service Providers Use Large Language Models in Workflows

Are language service providers (LSPs) seizing the benefits offered by generative AI and large language models? An April 2024 Slator study has found that the answer is a resounding “yes”.

The survey of AI adoption across 223 companies found that one in three LSPs have implemented large language models into their workflows. Furthermore, close to 20% of companies have now tested use cases or beta versions. 

Just 3% of LSPs told Slator that they have no plans to implement LLMs. 

Beyond performing the core machine translation task, LLMs — such as Meta’s Llama 3 and OpenAI’s GPT 4 and 4o — have significant potential to augment and enable language AI services. Key applications include quality estimation, dynamic translation adaptation, and multilingual text generation, among others.

Gabriel Fairman, CEO of Bureau Works, speaking at the opening of SlatorCon London on Wednesday 21 May, emphasized that LLMs represent nothing less than “a new way for humans to relate to machines”.

Language service providers are, consequently, moving fast to implement LLMs — typically via translation management systems (TMS) — to optimize processes and expand their service offering.

Adoption has, unsurprisingly, been most rapid across large LSPs. Slator’s survey found that 75% of Super Agencies have implemented LLMs compared to just under 30% of Boutiques.

Find the full survey results and detailed analysis in Slator’s 2024 Language Industry Market Report — Language AI Edition. The report includes a deep dive into LSP AI services strategy and market repositioning in the AI era.