For those who are passionate about language technology. Like us. Translated.

Milky Way
Joined March 2007
Does regulation hold innovation back, or can it help it succeed? In her latest article for Imminent, @RosariaTaddeo Taddeo, Professor of Digital Ethics and Defence Technologies at the @UniofOxford, challenges the idea that Europe must choose between technological progress and effective governance. She argues that the right regulatory frameworks can help Physical AI become safer, more reliable, and ready for widespread adoption. Discover why governance could be one of Europeโ€™s greatest competitive advantages. Read the article:go.translated.com/TaddeoInnoโ€ฆ
59
๐—ช๐—ต๐—ผ ๐—ฐ๐—ผ๐—ป๐˜๐—ฟ๐—ผ๐—น๐˜€ ๐—ฎ๐—ป ๐—”๐—œ ๐˜€๐˜†๐˜€๐˜๐—ฒ๐—บ ๐—ฎ๐—ณ๐˜๐—ฒ๐—ฟ ๐—ถ๐˜'๐˜€ ๐—ฏ๐—ฒ๐—ฒ๐—ป ๐—ฑ๐—ฒ๐—ฝ๐—น๐—ผ๐˜†๐—ฒ๐—ฑ? Across Africa, that question is becoming the real test of AI governance โ€” more consequential than any ethics statement or adoption target. Rwanda, Kenya, Nigeria, Morocco, and South Africa are treating procurement itself as a governance tool: using purchasing terms to demand data control, audit rights, and the authority to intervene once a system is running inside a public institution. In several respects, that puts them ahead of the United States, which still has no unified federal AI framework. The approaches differ across countries, Rwanda has paired AI adoption in healthcare and agriculture with investment in domestic technical capacity. Kenya has built on its "Silicon Savannah" ecosystem and the reach of M-PESA Africa to position itself in fintech and digital identity. Morocco is positioning itself as a multilingual governance hub between Africa, Europe, and the Middle East. Nigeria's startups are building language and financial-inclusion tools for markets larger platforms tend to overlook. South Africa is anchoring its AI debate in constitutional and human rights traditions. What ties these efforts together is a deeper question of autonomy rooted in language. Systems trained primarily on Western languages and categories can struggle to understand places like Kenya, where more than fifty languages and diverse cultural traditions coexist. Governance, in this sense, begins with understanding a place on its own terms. Arthur Sidney, Public Policy Strategist & Attorney, dives deep into the African ecosystem and what it reveals about who gets to shape AI governance next. go.translated.com/sidney26x
248
๐Ÿ—ฃ๏ธ "Language is the most profound manifestation of human evolution, and we believe that solving translation will open a new era of global collaboration." Those words from our co-founder Marco Trombetti set the tone for "AI, From the Inside", the event we hosted at Pi Campus on Monday, September 21, as part of @romefutureweek 2026. The co-founder Isabelle Andrieu retraced the story of Translated, we have been working on language and AI since 1999, back when teaching machines to grasp meaning still sounded like science fiction. Marco then looked ahead in his keynote, "What Comes After AI?" He shared what our long-term research tells us about how close AI has come to human-level quality in translation, and why scaling today's models will not be enough to close the remaining gap. Closing it calls for a different route, and that is where Lara 3, the latest version of our specialized translation AI, comes in, with an innovative model that learns by doing. It all started with a question that still drives Translated today: how can everyone understand and be understood in their own language? #AI #Translation #RomeFutureWeek
1
3
308
๐—ข๐—ป๐—ฒ ๐—บ๐—ผ๐—ป๐˜๐—ต ๐˜๐—ผ ๐—ด๐—ผ. The deadline to apply for the Imminent Research Grants is just one month away. Weโ€™re looking for researchers, innovators, startups, university labs, organisations, and companies working on the future of language research. Five projects will receive $20,000 each to support one year of research across areas including Language Data, Language Economics, Neuroscience of Language, Human-Computer Interaction, and Machine Learning for Translation. If you have an idea that could help us better understand where language and AI are heading, now is the time to apply. Deadline: 15 October 2026. Apply now and join a growing community exploring what comes next. go.translated.com/grants1monโ€ฆ
1
1
2
451
Translated. retweeted
Read here ๐Ÿ‘‰ slator.ch/TranslatedCEOLanguโ€ฆ At #SlatorCon San Francisco ๐ŸŒ‰ 2026, @Translation CEO Marco Trombetti spoke in the current limits to ๐Ÿค– #AI #translation, proposing three limitations that need to be addressed.
1
2
1
206
#Translation gives #AI nowhere to hide. Tomorrow at #SlatorCon San Francisco, our CEO Marco Trombetti will explore how translation reveals the limits of todayโ€™s AI architectures earlier than most applications. You will learn why recent improvements are slowing even as models grow larger and more expensive, and what research directions can address these limits, from parallel reasoning to learning by experience. Join us at 2:00 PM for ๐˜›๐˜ณ๐˜ข๐˜ฏ๐˜ด๐˜ญ๐˜ข๐˜ต๐˜ช๐˜ฐ๐˜ฏ ๐˜ข๐˜ฏ๐˜ฅ ๐˜ต๐˜ฉ๐˜ฆ ๐˜“๐˜ช๐˜ฎ๐˜ช๐˜ต๐˜ด ๐˜ฐ๐˜ง ๐˜Š๐˜ถ๐˜ณ๐˜ณ๐˜ฆ๐˜ฏ๐˜ต ๐˜ˆ๐˜ ๐˜ˆ๐˜ณ๐˜ค๐˜ฉ๐˜ช๐˜ต๐˜ฆ๐˜ค๐˜ต๐˜ถ๐˜ณ๐˜ฆ. #t9n #l10n #MT #localization
1
1
207
AI is good at learning from data. But human intelligence is shaped by something more fundamental: our ability to act, interact, make mistakes, and learn from what happens next. In our latest interview, Prof. A. Aldo Faisal explores the idea of embodied AIโ€”systems that operate within real-world behavioral loops, connecting perception, action, and feedback. His work looks at how understanding human behavior, from movement and gaze to everyday interactions, could help us build more capable and more responsive systems. From reinforcement learning to โ€œliving labsโ€ that capture natural human behavior, Faisal argues that the body offers important clues about how intelligence worksโ€”and how technology can be designed to assist people rather than replace them. This conversation is part of Imminent's editorial journey through the @dvps_ai project, exploring the technologies, infrastructures, and people shaping the future of hashtag#EuropeanAI. Read the interview here: go.translated.com/AIandthehuโ€ฆ
227
#Translation #AI is entering the age of ๐—ฒ๐˜…๐—ฝ๐—ฒ๐—ฟ๐—ถ๐—ฒ๐—ป๐—ฐ๐—ฒ. Weโ€™re bringing it to San Francisco. On September 3, Translated will be at #SlatorCon. Our CEO @marcotrombetti will speak on โ€œTranslation and the Limits of Current AI Architecture.โ€ Meet Michael Stevens, Dana Barras Franco, and Chris Thomas at our booth to discover Lara 3, a first-of-its-kind translation AI built on experience through Learning by Doing, and TranslationOS, our adaptive AI service delivery platform that orchestrates Lara and language professionals while automating localization from request to delivery. Attending? Reach out to meet us there. #TranslationAI #t9n #l10n #Localization
2
189
The case for ๐—ฝ๐—ฟ๐—ผ๐—ฝ๐—ฟ๐—ถ๐—ฒ๐˜๐—ฎ๐—ฟ๐˜† ๐˜๐—ฒ๐—ฐ๐—ต๐—ป๐—ผ๐—น๐—ผ๐—ด๐˜† in #localization? It's all about owning the conditions. Owning the inputs brings predictability. Owning the system brings control. Owning the pipeline brings transparency. None of it retrofits onto a foundation that wasn't built for it. Those who understand this will be better positioned tomorrow than the ones who picked the flashiest model today. Learn more about building vs. buying in localization in our latest report. Link in the first comment. #translation #t9n #l10n
1
1
267
๐—ช๐—ฒ๐—ฎ๐˜๐—ต๐—ฒ๐—ฟ ๐—ถ๐˜€ ๐˜๐—ต๐—ฒ ๐—ผ๐—น๐—ฑ๐—ฒ๐˜€๐˜ ๐—น๐—ฎ๐—ป๐—ด๐˜‚๐—ฎ๐—ด๐—ฒ ๐—ต๐˜‚๐—บ๐—ฎ๐—ป๐—ถ๐˜๐˜† ๐—ต๐—ฎ๐˜€ ๐—ฒ๐˜ƒ๐—ฒ๐—ฟ ๐˜๐—ฟ๐—ถ๐—ฒ๐—ฑ ๐˜๐—ผ ๐—ฑ๐—ฒ๐—ฐ๐—ถ๐—ฝ๐—ต๐—ฒ๐—ฟ. ๐—ฅ๐—ฒ๐—ฎ๐—ฑ๐—ถ๐—ป๐—ด ๐—ถ๐˜ ๐—ต๐—ฎ๐˜€ ๐—ฎ๐—น๐˜„๐—ฎ๐˜†๐˜€ ๐—ฏ๐—ฒ๐—ฒ๐—ป ๐—ฎ ๐—บ๐—ฎ๐˜๐˜๐—ฒ๐—ฟ ๐—ผ๐—ณ ๐˜€๐˜‚๐—ฟ๐˜ƒ๐—ถ๐˜ƒ๐—ฎ๐—น. But what happens when the grammar of that language changes? Weather AI is turning decades of scientific knowledge into something faster, cheaper and more widely available, changing who gets to read the weather and what we can do with that knowledge. And that matters. Weather shapes food, energy, health, infrastructure and economies, while climate change is making the cost of getting it wrong harder to ignore. The real leap is not only predicting weather better. It's making weather intelligence accessible: turning forecasts into decisions that help people and societies prepare, adapt and act. We spoke with Richard Turner, Research Lead for AI in Weather Prediction at Alan Turing Institute and Professor of Machine Learning at University of Cambridge, about how AI is reshaping the way we understand weather, and what comes next. Read the full interview on Imminent. go.translated.com/richardturโ€ฆ
244
๐Ÿค” Every breakthrough in hashtag#AI begins when the way it learns changes. We have moved from rules to data to prediction. What comes next? Join us today to explore what ๐—น๐—ฒ๐—ฎ๐—ฟ๐—ป๐—ถ๐—ป๐—ด ๐—ณ๐—ฟ๐—ผ๐—บ ๐—ฒ๐˜…๐—ฝ๐—ฒ๐—ฟ๐—ถ๐—ฒ๐—ป๐—ฐ๐—ฒ means for #translation AI and introduce the next chapter of Lara and Lara Translate. โœจ #t9n #l10n go.translated.com/lara3
202
AI can beat any human at chess, but it still can't fix a leaking pipe. For Alexander Waibel, professor at @CarnegieMellon and the Karlsruhe Institute of Technology (KIT), that gap reveals more about intelligence than any benchmark ever could. For Waibel, intelligence goes far beyond scaling learned information. It lies in the ability to understand context and apply knowledge differently from one situation to the next. It's a perspective that suggests the next breakthrough in AI won't be driven by more data, but by deeper contextual understanding. Read the interview with the pioneer of machine translation to explore where the next frontier of artificial intelligence is heading. go.translated.com/Scientificโ€ฆ
242
๐Ÿ“–ย ๐—ข๐—ฝ๐—ฎ๐—ฐ๐—ถ๐˜๐˜† ๐˜๐—ฎ๐˜… (๐˜ฏ๐˜ฐ๐˜ถ๐˜ฏ) The hidden cost of building an AI workflow from multiple models, providers, and integrations that you cannot fully see or control. Symptoms include: - Unexpected model changes. - Inconsistent quality. - Unclear accountability. - More time spent managing the stack than improving outcomes. Example: ๐˜›๐˜ฉ๐˜ฆ ๐˜—๐˜ฐ๐˜ณ๐˜ต๐˜ถ๐˜จ๐˜ถ๐˜ฆ๐˜ด๐˜ฆ ๐˜ธ๐˜ฐ๐˜ณ๐˜ฌ๐˜ง๐˜ญ๐˜ฐ๐˜ธ ๐˜ง๐˜ข๐˜ช๐˜ญ๐˜ฆ๐˜ฅ, ๐˜ฃ๐˜ถ๐˜ต ๐˜ฏ๐˜ฐ ๐˜ฐ๐˜ฏ๐˜ฆ ๐˜ค๐˜ฐ๐˜ถ๐˜ญ๐˜ฅ ๐˜ช๐˜ฅ๐˜ฆ๐˜ฏ๐˜ต๐˜ช๐˜ง๐˜บ ๐˜ธ๐˜ฉ๐˜ฆ๐˜ณ๐˜ฆ ๐˜ช๐˜ฏ ๐˜ต๐˜ฉ๐˜ฆ ๐˜ฑ๐˜ช๐˜ฑ๐˜ฆ๐˜ญ๐˜ช๐˜ฏ๐˜ฆ ๐˜ช๐˜ต ๐˜ฉ๐˜ข๐˜ฑ๐˜ฑ๐˜ฆ๐˜ฏ๐˜ฆ๐˜ฅ, ๐˜ธ๐˜ฉ๐˜บ ๐˜ช๐˜ต ๐˜ฉ๐˜ข๐˜ฑ๐˜ฑ๐˜ฆ๐˜ฏ๐˜ฆ๐˜ฅ, ๐˜ฐ๐˜ณ ๐˜ฉ๐˜ฐ๐˜ธ ๐˜ต๐˜ฐ ๐˜ด๐˜ต๐˜ฐ๐˜ฑ ๐˜ต๐˜ฉ๐˜ฆ ๐˜ด๐˜ข๐˜ฎ๐˜ฆ ๐˜ช๐˜ด๐˜ด๐˜ถ๐˜ฆ ๐˜ง๐˜ณ๐˜ฐ๐˜ฎ ๐˜ณ๐˜ฆ๐˜ข๐˜ค๐˜ฉ๐˜ช๐˜ฏ๐˜จ ๐˜ต๐˜ฉ๐˜ฆ ๐˜ฏ๐˜ฆ๐˜น๐˜ต ๐˜ฎ๐˜ข๐˜ณ๐˜ฌ๐˜ฆ๐˜ต. ๐Ÿ‘‰ Learn how to recognize the opacity tax in our Build vs Buy report. โฌ‡๏ธ Link in the first reply. โฌ‡๏ธ #t9n #l10n
2
1
222
๐Ÿ“How old is your AI, and how much #translation experience does it have? ๐Ÿ‘‡ Join us on July 30 to discover why ๐—ฒ๐˜…๐—ฝ๐—ฒ๐—ฟ๐—ถ๐—ฒ๐—ป๐—ฐ๐—ฒ will drive the next leap in #AI and what that means for the future of translation. Weโ€™ll also unveil the latest version of Lara and introduce a new @LaraTranslate offering designed to better meet the needs of todayโ€™s workplace. #t9n #l10n #localization hubs.la/Q04qw6wn0
140