University of Southern California’s Post

"The reality is that, despite advances in AI, jobs for human interpreters and translators are not cratering. In fact, the data suggests they’re growing." “Companies continue to hire human translators and interpreters because they understand that AI cannot replace the expertise and judgment that these professionals bring to the table. This is particularly true for high-stakes projects in fields like legal, medical, but also literary translation, where accuracy and cultural appropriateness are paramount.” Thank you Bridget Hylak, CI, CT, MTC and Daniel Sebesta for representing the profession in this great article! It's the kind of content I want to see more often: Less hype and more reality. 👏

🔥Effectiveness - accuracy, relevance and contextuality among other requirements - is the name of the game in language management. So, this announcement sounds like good news and food for testing🔥 🎇In my experience DeepL has a history of investing significantly in human language expertise and experience to build and improve their models. The value of this differentiator has come up clearly in some industries (life science, legal, finance) and in some content (narrative, descriptive) I had the chance to test and evaluate. 🌟The other benefit of such a specialized large language model is to make generative AI (more) inclusive by potentially enabling (more) business users and functions to leverage in light of the delivered effectiveness. 🔎I can't wait to put this one to the test.

https://lnkd.in/eqJZ2hgi What do you think?? Leveraging advanced language models like GPT-4 for this task could help detect and correct errors in machine-generated translations, enhancing their reliability. “LLM based automatic translation post-editing could aid in both detecting and fixing translation errors to ensure greater reliability of MT outputs,” the researchers said. They also underlined that LLMs’ multilingual capabilities make them suitable for automatic post-editing tasks because beyond error correction they can also apply knowledge-based or culture-specific customizations to translations.

Machine translation technology has evolved significantly in recent years, allowing for more accurate and natural language translations

🌐💬 Ever wondered how language translation apps seamlessly convert your words into another language? What's the linguistic sorcery that bridges communication gaps effortlessly? 1. Neural Machine Translation (NMT): The core engine behind these apps is often powered by Neural Machine Translation. Think of it as an AI language maestro. It analyzes vast datasets to understand the nuances and context of different languages. 2. Contextual Understanding: These apps don't just do word-for-word swaps. They grasp the context of your sentences. This ensures translations capture the intended meaning, not just the literal words. 3. Constant Learning: Like a language scholar on a perpetual quest for knowledge, translation apps learn as they go. The more you use them, the better they get at understanding your unique style and preferences. 4. Multilingual Magic: Some apps can juggle multiple languages in one conversation. They can seamlessly switch gears between languages, making multilingual communication a breeze. 5. Real-time Adaptation: Need to chat with someone in real-time? No problem. Translation apps work on-the-fly, rapidly adapting to the conversation's pace and nuances. 6. Cultural Sensitivity: These apps aren't just about words; they're culturally savvy. They recognize idioms, colloquialisms, and regional variations, ensuring your message transcends language borders. 7. Speech-to-Text Brilliance: Many apps allow you to speak, and they transcribe and translate in real-time. It's like having a personal interpreter in your pocket, ready to facilitate global conversations. Follow for more #simpleandfascinating

BigTranslate:  Unlocking 100+ Language Translation with Large Language Models ... 🧠 I am excited to share an intriguing new research paper from the Institute of Automation at the Chinese Academy of Sciences titled "BigTranslate: Augmenting Large Language Models with Multilingual Translation Capability over 100 Languages." As many of you know, I have a strong interest in advancing natural language processing and democratizing access to AI. 👉 This paper tackles two key challenges: 1 - most large language models today are English-dominant and only support translation between a few dozen language 2 - there is an imbalance between high-resource and low-resource languages, making translation quality inconsistent. 👉 The researchers present BigTranslate, which enhances the LLaMA-13B foundation model to translate over 100 languages. The key steps: - Pre-train LLaMA on 42.8B Chinese tokens to improve Chinese language mastery - Continue pre-training on 47B parallel tokens covering 102 languages, using curriculum learning to balance high-resource and low-resource languages - Fine-tune with 240K multilingual translation instruction examples The result is a model that can translate 102 languages at a level comparable to ChatGPT and even surpassing it for 8 language pairs. For instance, English→Amharic translation achieves 31 BLEU score compared to 17.3 for ChatGPT. While this research is still preliminary, I am thrilled by the possibilities. Massively multilingual models like BigTranslate could one day provide high-quality translation to billions of people in their native languages. The techniques used here also demonstrate how we can continue to improve access to AI for lower-resourced languages. I encourage you all to check out the full paper and let me know your thoughts! What other creative ways can we build fairer, more inclusive AI systems?

ARTIFICIAL INTELLIGENCE: A THREAT FOR LINGUISTS OR AN OPPORTUNITY? Many linguists have been worrying about the alleged threat posed by AI to language professionals. Although I am absolutely positive that AI will reshape the way we language professionals work, I think the need to learn foreign languages will never cease to exist. ✅ some translation and interpreting work will always be done by experienced linguists - no matter how flawless AI tools may be, linguists are liable for their performance, whereas machines are just machines and their output will always be scrutinized by human experts, not the other way around. ✅ language skills are useful in many professions; sure, you can talk to your business partner using machine translation and interpreting, but what about your credibility? Why not go the extra mile and learn - say - Korean to talk to your prospective customer in Seoul? ✅ Reaserch, research, research! Research in humanities has always existed and linguistics is no different than other areas, there will always be people analysing the subtleties of languages and their history. ✅ The pleasure of learning a foreign language: we should not forget that learning foreign languages is not just something you do for your future profession, but rather because it is a funding moment in your personal development. What do you think about that? What will be the impact of AI on language professionals and language learning in general? #translation #interpreting #languagelearning #languageteaching #AI https://lnkd.in/eZfYBrek

🌍 Giving a Voice to the Unheard: Multi-Language Support for Least Spoken Languages in LLMs In a world rich with linguistic diversity, many languages remain underrepresented in digital spaces. At ProCogia, we're changing that narrative! Our latest blog explores how leveraging closed-source machine translation layers with large language models (LLMs) can enhance communication across South African languages like Zulu, Xhosa, Northern Sotho, and Afrikaans. 🔗 Dive into our findings and see how we're making strides towards more inclusive AI development. Let's ensure every language has a voice in the future of technology! Read the full article here 👉 https://lnkd.in/g9HtEhc2 #LanguageModels #AI #Inclusion #Technology #ProCogia