Computational linguistics has evolved in amazing and impressive ways. With the advent of Generative AI, the world has been exposed like never before to what can be done by making large corpora of language accessible to increasingly powerful compute environments. Nevertheless, there is no real "intelligence" in AI, and language evolves in ways that are only partially understood, especially in a multilingual context. In this session, Dr. Anthony Scriffignano will discuss the evolution of the datasphere (information available to computers) and specific confounding characteristics of language that make it challenging for AI. This session will also focus on common myths and inconvenient truths that make multilingual semantics an ongoing challenge in the real world with examples from business and real-world anecdotes.

NVIDIA is one of the most important companies driving the revolution in the computer industry, as generative AI techniques like Large Language Models enable  modern AI systems to perform tasks previously considered something only humans could tackle.  In his role as VP of Applied Research at NVIDIA, Jonathan Cohen oversees the development of the NeMo platform for developing, deploying, and improving Large Language Models and other conversational AI systems.  This fireside chat will focus on the key question of this conference: in what ways do the AI and Language industries need each other?  AI systems are powered by compute, data, and algorithms.  But in practice, to create and deploy them into real production settings requires overcoming significant complexities beyond just these 3 ingredients.

In this fireside chat with Jonathan Cohen, we will discuss what are the real challenges facing the AI industry, and how can the skills and data brought by the translation and language industry help solve these problems.  As AI systems are able to perform more automation and demonstrate “intelligence” that was previously reserved for humans, what are the new opportunities available to service providers and individual experts in the language industry?  Conversely, how can modern AI systems be used effectively by the language industry to solve some of their most challenging problems?  In what ways are these two industries complementary, and how should we think about their convergence?

The idea for this session is to give the audience insights into the thinking and perspectives on AI from the Big Tech research teams that are responsible for the largest and most influential models, and that have already integrated AI into their businesses despite the risks and disruption to their existing business models that AI may pose.

DeepL has stayed pretty much under the radar of the localization industry ever since the company was founded in 2017. Increasingly the name came up in online forums and at industry conferences. DeepL has made it to the top-ranking MT engines in quality benchmarking reports. But the people and the faces behind this success remained hidden. Until now. At the TAUS conference in Albuquerque Jarek Kutylowski, founder and CEO of DeepL, will share the story of DeepL from the origins of Linguee to the recent 2 billion valuation of the company. And what does that mean for the future of DeepL (and for the world)?

Nine LSPs and start-ups are selected to pitch their AI solution in six minutes. The audience will choose the winner of the TAUS Massively Multilingual Contest.

Globalization leaders and CTOs have no other choice than embracing AI. How do they go about it and what is it going to bring them in return? Should translation buyers own their workflows and get their AI independent of the TMS and LSP?