Macduff Hughes spent 12 years leading the Google Translate team at Google — guiding it through two of the most consequential technology shifts in the history of language. In this conversation he breaks down how Google Translate actually worked under the hood, what the 2016 neural transition really looked like from inside the room, and what the rise of LLMs means for the future of translation and the professionals who work in it.
Macduff Hughes spent 35 years in tech — including 8 years managing the Adobe Acrobat team and 12 years leading Google Translate. He co-authored the landmark 2016 GNMT research paper alongside Jeff Dean and 29 others, overseeing an error reduction of up to 85% on some language pairs overnight. He retired in 2024 having guided the team through statistical MT, neural MT, and into the LLM era.
In this episode:
- How statistical machine translation actually worked — and why it had a ceiling
- The 2016 neural transition: the decision, the disruption, and the human side
- What 500 million users and 200 languages taught him about language access and equity
- Gender bias in MT training data — and how the team tried to address it
- What the LLM era means for dedicated translation products
- Honest advice for mid-career translators navigating the shift
Watch on YouTube: https://youtu.be/dwYgYj_Cmvg
Connect with Macduff Hughes: https://www.linkedin.com/in/macduff-hughes-98665a3
Localization Fireside Chat on LinkedIn: https://www.linkedin.com/company/localization-fireside-chat
Robin Ayoub on LinkedIn: https://www.linkedin.com/in/robinayoub
Blog: https://www.robinayoub.blog
N49Networks: https://www.n49networks.com
Subscribe on YouTube: https://www.youtube.com/@L10nFiresidechat
Book a call with Robin: https://calendly.com/robin-ayoub/30min