| Mobile Web

Interpretation and translation battleground is localisation, competing with frontier models with on-site data

Flitto co-founder and CTO Kang Dong-han said generic AI is hard to apply directly at industrial sites because accents, dialects, workplace jargon and machine noise affect performance. He said interpretation and translation quality depends on how quickly AI is localised for the site, focusing on fine-tuning frontier models with on-site data and glossaries. Flitto built an AI interpretation and translation service for Hanwha Ocean and Hanwha Systems, supporting 44 languages and functions such as group chats and image translation.