Celebrating European Translation Technology

AI seems to make everything so easy. Heads of enterprise globalization and localization get the question asked from their executives: Why do we need a localization vendor? Can we not just use a LLM? 

It is tempting for enterprises to build their own multilingual content delivery platforms on the back of GenAI. Do-It-Yourself is what chiefs of technology may articulate as the new localization strategy. And save all the costs spent on human translation and quality assurance. The heads of the localization departments tend to resist this oversimplified approach. But at the same time, they know that drastic change is inevitable.

This session is split in two panel discussions: pragmatists and radicalists. The pragmatists have an evolutionary approach to AI technology: they adapt and make changes step-by-step. The radicalists are supporters of the transformation approach, wiping out existing tools, habits and processes, and going all-in with AI technology.

The two panels each with opposite perspectives will give the audience a deep understanding of the challenges and dilemmas customers face at both sides of the equation. Synergies between the different strategies will be sought in a conclusive Q&A session between the moderators and the audience.

In the EU Localization Championship, each company gets exactly ten minutes to present their hands-on solution. The audience will vote for the winners at the end of all the presentations.

What are the synergies and divisions in the globalization strategies?

Human intelligence remains indispensable, but translation services are changing. Language Service Providers need to reinvent themselves. Will they move up the value chain, become tech-first companies? Or will they go horizontally, leveraging their networks of human linguistic resources to offer data services: the human data layer.

Moving up the value chain and becoming a tech-first global content service provider (GCSP) seems an attractive strategy, but also brings along dilemmas such as build-or-buy decisions.

Going horizontal and redirecting the translation resources to data services - annotating data, validating data, quality review - is another strategy that LSPs are following. Is this an effective strategy for reinventing the LSP business or is it a race-to-the-bottom with prices tumbling down and competition becoming ever more severe?

Long before the private sector dared to rely on automatic translation the European Commission already started experimenting with the technology from Systran. The history of MT at European institutions goes back to the nineteen-seventies and has since then pioneered every wave of innovation. European institutions, as big users of translation technology and the largest employers of translation professionals, have a wealth of experience and insights to share.

In this session we will hear from the leaders of technology, quality and human resources at government institutions in Europe about their past, present and future experiences and plans with translation technology, and AI.

MT is a given. Every institution and business is using it. The next AI technology breakthrough is Quality Estimation (QE) and Automatic Post-Editing (APE). A recent Gartner report on QE and TAUS predicts that by 2027 all translation operators in the world will have implemented QE. And as Nimdzi wrote last year: “If nearly 50% of the market is using MTPE, but not all of them are leveraging the full power of MTQE and AutoPE, then we’re looking at systemic inefficiency at scale. Even a conservative estimate suggests billions of dollars are spent globally on partly avoidable human labor, operational drag, and underutilized automation.”

In this session we talk with both enterprise and LSP users about their experiences with risks and confidence analysis and the results they are able to achieve.

In the current landscape of AI-driven content generation, the industry has focused extensively on speed and scale. Less time has been spent on interrogating the interactional layer of this new delivery paradigm: how AI-generated multilingual content positions itself in dialogue with users - in this case, content authors in the enterprise - and how that positioning shapes trust, uptake, and decision-making. As large language models become more pervasive in global content workflows, they do far more than translate. They actively perform alignment work to reduce friction, drive productivity, and lower costs. But at what future cost? 

This discussion will explore the risks of processing multilingual content through LLMs and cover the critical "downstream" effects of this dynamic: the masking of uncertainty and inaccuracy through high fluency, potentially leading to quality concerns and the erosion of brand integrity.