You know when you go to an event and bring a book that you plan to read but forget about it? Today, I pulled out the Imminent Report from Translated that was distributed at the TAUS Conference in Rome and started reading the article "The Path to LLM-based Machine Translation" by Kirti Vashee and couldn't put it down. Then I read the story about how "The Human Touch" video - that makes me cry every time I see it - was made and all the cultural nuances of gestures and non-verbal communication by Patrizia Boglione. There are so many great articles that I must read still. And what a beautiful production! If you want a copy, you can get it for free here: https://lnkd.in/gWnMghky. You won't regret it.
Translated just released their 2024 Imminent Research Report, "Symbiotic Connections." This year's edition goes into the relationship between humans and machines. It explores how AI is influencing language and cognitive processes, the potential of large language models for revolutionizing translation, and the challenges and opportunities that arise when designing socio-technical systems. Read more: https://buff.ly/3xPWKkW #Translated #MultiLingualMedia
Terence Lewis
8 个月
I too had tears in my eyes when I watched that video. Blokes can allow themselves the odd tear!
Daniel Sebesta
8 个月
Oh yes, that was a wonderful item to take home from the conference. Thanks for the reminder, Renato—I too was planning to read a few more articles!
Kirti Vashee
8 个月
While the article has my name on it, several people at Translated made significant contributions to it to make it a more complete presentation
Experienced freelance translator
8 个月As far as Translated, aka Modern MT is concerned, there are actually three things that make me want to cry: 1. The fact that they expect me to work for 70% of my rate, despite selling their product for <1% of it. 2. The fact that about the same reduction is expected by my outsourcer who apparently uses some "proprietary" version of this contraption, despite the fact that the quality of the output is 3. another thing that makes my want to cry. Even the basic ambiguities aren't rendered properly, which results in absolute nonsense provided in a bit too many places in which there is even a slight possibility of getting something really wrong. Moreover, as far as the famous "learning" feature is concerned, well, the AI it wasn't even able to get the NAME of the machine trom the existing TM properly, not to mention keeping such trifles as style or grammar consistent.