DeepL and AI in Translations
AI applied to translations is entering a new phase with DeepL: no longer just a tool for end users, but a critical infrastructure for governments, courts, and large organizations. This transformation is highlighted by the CEO of DeepL, Jarek Kutylowski, in a Q&A at HumanX, where the tensions between technological innovation, shortage of human resources, and regulation clearly emerge.
AI and Translations: When Technology Replaces the Lack of Interpreters
One of the most tangible signs of the growing adoption of AI in translations comes from the public sector. DeepL is indeed in negotiations with the UK judicial system for a potential collaboration. The issue at hand is not the cost of interpreters, but their availability. According to Kutylowski, “There aren’t enough qualified interpreters. It’s a human resources issue, not an economic one.”
This scenario is echoed globally, as DeepL is already collaborating with governments and institutions in countries such as Canada, Switzerland, Japan, and the United States (particularly California), where the need to translate documents into multiple languages has become structural. In multilingual contexts, AI is no longer an option but an operational necessity, explaining the rapid growth in demand precisely in the most linguistically fragmented areas.
Europe vs USA: AI Regulation Risks Hindering Innovation
The most controversial issue surrounding AI in translations is the regulation of artificial intelligence. Kutylowski takes a clear stance: an excess of rules could have counterproductive effects. He states, “If regulation becomes too restrictive, as a founder, I would move to the United States.” This concern indicates a genuine risk of companies and talent migrating towards more permissive ecosystems.
European startups could find themselves at a competitive disadvantage compared to their American counterparts, which operate under fewer regulatory constraints. This sentiment is echoed by Victor Riparbelli, CEO of Synthesia, who emphasizes how London and Europe risk losing ground due to difficult-to-implement regulations. The message is clear: some regulation can encourage adoption, but an excess risks stifling innovation.
Global Markets: Between Asian Enthusiasm and European Caution
The adoption of AI for translations is not uniform globally. According to DeepL, Europe and Asia currently represent the markets with the highest number of users, yet they demonstrate very different attitudes. In Europe, the market is still relatively cautious. Companies are adopting AI, but often hindered by legal and compliance issues.
Conversely, Asia exhibits a decidedly more enthusiastic approach, even though countries like Japan have longer and more complex internal processes for technological implementation. In the United States, the context differs: the widespread use of English reduces the need for translation tools, while fostering a more dynamic environment for the development of AI companies.
The Real Limitation of AI? Data (and “Difficult” Languages)
One of the less discussed yet critical aspects of AI in translations concerns data availability. Effective translation models require vast amounts of linguistic content. Kutylowski notes, “We are always looking for new languages to add, but it depends on the availability of data to train the models.” This creates a significant gap, especially for some African or minority languages, where datasets remain insufficient. The result is incomplete global linguistic coverage, risking further widening the digital divide.
Events and Operational Limits: Why AI Doesn’t Replace Everything (Yet)
Despite advancements, AI is not yet the perfect solution in all contexts. For instance, DeepL does not focus significantly on translations for live events. Practical challenges such as logistics, headsets, and technical complexity make adoption less efficient compared to other areas. This underscores a fundamental point: artificial intelligence excels when it can operate on a large scale and in structured environments, yet still encounters limitations in highly dynamic contexts.
The Future of AI Translations: Global Infrastructure (but Regulated?)
The DeepL case demonstrates how AI for translations is becoming an infrastructural technology, increasingly integrated into public and corporate processes. However, the future hinges on a delicate balance: the need to regulate for security and trust against the tangible risk of losing global competitiveness.
If Europe fails to find this balance, as suggested by the CEO, the next generation of AI companies may not emerge—or remain—on the continent. At that point, the game of artificial intelligence will be played elsewhere.
