The language industry is, of course, well ahead of the curve when it comes to AI adoption: machine translation is at the heart of the industry’s core workflow. No surprise then, that LSPs are actively seizing the benefits offered by generative AI and large language models (LLMs).
According to an April 2023 study by Slator, one in ten LSPs have already run pilots or developed beta integrations that apply the powerful capabilities of LLMs. Furthermore, more than 90% plan to apply LLMs in their business.
The survey results — which are provided in full in Slator’s flagship annual Language Industry Market Report for 2023 — are based on responses to questions on AI adoption from more than 50 CEOs (and other C-level roles) at leading LSPs in Slator’s LSPI.
The most promising LLM applications for the industry, according to LSPs, are in machine translation, quality evaluation and business workflows.
Early movers include Unbabel — which has developed an LLM-enabled quality evaluation tool — and Smartling, which has designed a (patent pending) prompt engineering process. Lilt, meanwhile, has built an LLM that applies in-context learning to customize outputs for legal and marketing use cases.
LSPs are also exploring the ability of LLMs to generate original content in multiple languages. It’s a new advance that is likely to drive both disruption and opportunities for industry players. Middle East-focused LSP Tarjama told Slator that the company is currently testing LLMs for content creation use cases.
And what about translation buyers? Will companies start to use ChatGPT to generate source texts and send such content to LSPs for translation? “It’s already happening”, according to one in four LSPs that took part in the survey.
Find the full survey results and detailed analysis in Slator’s 2023 Language Industry Market Report. The report includes a deep dive into generative AI and large language models — their key capabilities and applications, as well as their potential implications for language industry players.