Alibaba’s global unit pushes translation-focused AI model for merchants, individual users

Macro MT is a new large language model that supports translation in various cross-border e-commerce scenarios

Reading Time:2 minutes

Alibaba’s new translation-focused large language model can be integrated into merchants’ systems that manage their products, customer services and search guidance. Photo: Shutterstock

Ann Caoin Shanghai

Published: 9:45am, 16 Oct 2024

The international commerce arm of Alibaba Group Holding on Wednesday launched a self-developed large language model (LLM) – the technology underpinning generative artificial intelligence (GenAI) applications like ChatGPT – that is focused on translation.

The new AI model, Macro MT, was unveiled by Alibaba International Digital Commerce Group (AIDC) to support translations in various cross-border e-commerce scenarios and everyday communications, according to Zhang Kaifu, vice-president and head of AI initiative at AIDC. Alibaba owns the South China Morning Post.

“Macro MT will seamlessly integrate into Alibaba International’s overarching business strategies, fortifying the operational backbone of our diverse cross-border e-commerce platforms such as AliExpress, Lazada, Trendyol and more,” Zhang said in an interview on Tuesday.

For merchants, Macro MT can be integrated into systems that manage their products, customer services and search guidance. The LLM can translate merchant-uploaded product information including titles, descriptions and image captions into the language of target markets. It is also capable of translating search keywords and dialogue, according to the new LLM’s launch page.

Merchants on AliExpress, the international online shopping platform of Alibaba Group Holding, are expected to benefit from the new Macro MT large language model. Photo: Shutterstock

The model currently supports translation in 15 languages, including Chinese, English, French, German, Italian, Japanese and Korean.

“We plan to expand the number of languages supported, further enhance translation quality across a broader range of scenarios, and consistently reduce the cost of usage for our users,” Zhang said.