A deep dive into Translation Trends: Key takeaways from Tekom Stuttgart 2024

The latest edition of the Tekom conference, held in Stuttgart, Germany, highlighted a range of developments in language technology, with the focus on automated quality evaluation (AQE), post-editing (APE), and maintaining clean terminology management through collection, extraction, and enrichment. Tekom, or TCWorld Conference, brings together participants, speakers, students and exhibitors of "technical communication" from all over the world to the largest industry meeting place. 

Acolad was present as an exhibitor and also as a speaker in one of the panels, with Leena Peltomaa, Global Solutions Manager, speaking on “Revolutionizing DEI and Regulatory Compliance in Global Tech Communication”. 

Emerging trends: AI-Driven quality and responsible processes

AI remained a central theme, particularly regarding quality assurance and ethical responsibility in AI-supported processes and documentation. Automated quality evaluation and post-editing were widely discussed, with companies increasingly emphasizing the importance of clean terminology, Retrieval-Augmented Generation (RAG), and knowledge graphs to fine-tune AI outputs. Terminology-focused tools addressing collection, extraction, and enrichment were in demand as companies recognized the value of precise language resources for enhancing translation accuracy.

Navigating a complex solution landscape

While the conference introduced numerous AI and language tech solutions, many attendees expressed difficulty navigating the options. The rapidly evolving landscape poses a challenge for customers trying to understand and implement these solutions effectively. The proliferation of new tools underscores the need for clear guidance on how to integrate and make the most of AI-driven language technologies.

Across the board, competitors in the language tech space focused on AQE, APE, and clean terminology use, with innovative approaches to terminology collection, extraction, and enrichment emerging as critical themes. Companies are increasingly leveraging RAG and knowledge graphs to enhance AI accuracy and adaptability, aiming to ensure that AI outputs align with customer expectations and specific industry standards.

Acolad’s approach: Quality evaluation from multiple perspectives

Leena Peltomaa, Global Solution Manager at Acolad, spoke on the group’s comprehensive approach to quality evaluation - exploring quality assessment for both monolingual and translated content. Acolad's approach incorporates various perspectives—including diversity, equity, inclusion (DEI), regulatory compliance, and translation accuracy—to provide a robust framework for quality assurance and demonstrates how AI and LLMs offer rapid and easy quality overviews, as well as the ability to implement real-time adjustments to enhance content.

The management of AI’s limitations was a transversal topic to the event, especially around regulatory compliance, data privacy, and the sometimes-inconsistent quality of AI-generated content. Adopting a model-agnostic approach and using LLMs with tailored prompts and terminology emerged as key strategies to optimize AI’s effectiveness. Integrating AI tools that allow for customized outputs based on specific content needs was also highlighted as a way to address the unique demands of quality evaluation.

Translation tech evolution: Embedding real-time translation

Developers are increasingly embedding translation capabilities within applications—such as real-time, automated translations for customer support portals—often using LLMs like OpenAI’s, which excels at handling “noisy” content. However, neural machine translation (NMT) continues to outperform LLMs in specialized language tasks. While LLMs are transforming workflows, full integration into established business models remains a gradual process.