The State of Machine Translation 2023 Report Evaluates 37 MT Engines and 5 Large Language Models

Intento, the Enterprise Machine Translation Hub, and e2f, a language intelligence pioneer, teamed up to study the Machine Translation market and present the State of Machine Translation 2023 report (the seventh consecutive year for Intento).

This edition analyzes the performance of 37 MT Engines across 22 language pairs and 9 content domains, providing the latest overview of the rapidly evolving MT landscape. Key highlights include:

  • Out of the 37 Machine Translation Engines evaluated, 23 are best-performing in some language pairs and domains

  • 5 Large Language Models from AI21, Anthropic, and OpenAI, including ChatGPT and GPT-4, have risen to the first tier in MT capabilities

  • All of the 48 existing MT vendors cover 654 unique languages and 190,000 language pairs

  • Few MT vendors perform at the top level for Arabic, Japanese, and Ukrainian languages

  • Legal and IT domains require careful MT vendor selection, while Entertainment and Colloquial domains may require customization for optimal results

The report spotlights the MT landscape transformation with new features, expanded offerings, and language pairs from several vendors – all demonstrating significant progress in technology and innovation. Drawing from Intento’s expertise, businesses can use this shift to enhance language capabilities, improve global communication, and stay competitive.

It also identifies top-tier LLMs like ChatGPT and GPT-4, which surpass other models in various MT evaluation metrics and provides insight into the potential of GPT-3, Anthropic’s Claude, and AI21’s Jurassic-2 models for businesses exploring emerging language technologies.

We launched our inaugural MT report in 2017, at a time when there were only eleven stock models available. Fast forward six years, and the MT landscape has exploded, with six-fold growth in MT platforms and 15x more language pairs,” says Konstantin Savenkov, Intento CEO and co-founder. “In addition, we see new automated customization possibilities such as domain adaptation and custom glossaries, and the emergence of Large Language Models, which can handle translation among other tasks.
— Konstantin Savenkov, Intento CEO and co-founder
We’re excited to work with Intento, providing the best linguistic data to drive exceptional MT evaluation results in the latest report,” says Michel Lopez, CEO of e2f. He continues, “The human effort to create this linguistic data is essential to generating these golden datasets. Top AI companies who need the best language and data science expertise are increasingly turning to our workforce of linguists and data scientists as they grapple with the challenges of quality, cost and domain expertise in this era of generative AI.
— Michel Lopez, e2f CEO and founder

About Intento

Intento enables global companies to translate 20x more on the same budget and makes their customer and employee experience instantly multilingual in 650+ languages through Machine Translation and AI. 

Intento Enterprise MT Hub augments best-of-breed Custom NMT platforms with Source Quality Improvement, Automatic Post-Editing, and Translation Quality Estimation. Combining Machine Translation with Generative AI models (like GPT) enhances traditional translation workflows and achieves accurate, in-context translation.

It integrates with existing software platforms, such as Salesforce, ServiceNow, and Translation Management Systems, to share MT/AI models across the enterprise and improve them based on feedback.

For more information, visit https://inten.to or book a demo with an Intento expert.

About e2f

Established in 2004, e2f helps people and machines communicate naturally, regardless of language, content, and culture. e2f solutions empower Fortune 50 brands to monitor, objectively assess, and improve communications on a global scale.

e2f delivers world-class translation and training data with its proprietary technology stack for AI services, translation and quality review. e2f offers a global resource pool of skilled professionals in virtually all countries and languages.



Previous
Previous

Notes from the CEO: How the Language Industry can adapt and thrive in the Age of AI

Next
Next

Privacy First: 5 Challenges for AI-powered Chatbots