PeopleofSuperintelligence

Researcher

Jacob Devlin

Jacob Devlin is a researcher known for co-writing BERT, a bidirectional Transformer language model submitted to arXiv in 2018. He earned a master's degree in computer science at the University of Maryland in 2009, and a paper he co-wrote on neural network joint models won the 2014 Best Long Paper award from the Association for Computational Linguistics. From 2014 to 2017 he worked at Microsoft Research, where he led the move of Microsoft Translate to neural machine translation. In 2018 he and Ming-Wei Chang posted the open-source release of BERT on Google's research blog. He left Google for OpenAI in January 2023, and The Information reports that he has since returned to Google.

Timeline

  1. 2009

    A master's degree at Maryland

    He earns a master's degree in computer science at the University of Maryland in 2009.

  2. 2014

    An ACL Best Long Paper award

    His co-authored paper 'Fast and Robust Neural Network Joint Models for Statistical Machine Translation' wins the 2014 Best Long Paper award from the Association for Computational Linguistics.

  3. 2014–2017

    Microsoft Translate goes neural

    He works as a principal research scientist at Microsoft Research, where he takes Microsoft Translate from phrase-based translation to neural machine translation.

  4. 2018

    The BERT language model

    He co-writes the BERT paper, 'Pre-training of Deep Bidirectional Transformers for Language Understanding', submitted to arXiv in 2018. He and Ming-Wei Chang post the open-source release of BERT on Google's research blog.

  5. January 2023

    A move to OpenAI

    He leaves Google for OpenAI in January 2023, after complaining internally that Bard was being trained on ChatGPT data. Google responds that Bard is not trained on any data from ShareGPT or ChatGPT.

  6. Since 2023

    Back at Google

    He returns to Google after his time at OpenAI, as The Information reports.

Sources and credits

Photo: We have no published photo of Jacob Devlin yet. Send us a free-licensed photo.

  1. Jacob Devlin wins ACL Best Paper Award cs.umd.edu. Checked 14 October 2026.
  2. Stanford NLP Seminar: Jacob Devlin nlp.stanford.edu. Checked 14 October 2026.
  3. [1810.04805] BERT: Pre-training of Deep Bidirectional Transformers for Language Understanding arxiv.org. Checked 14 October 2026.
  4. Open Sourcing BERT: State-of-the-Art Pre-training for Natural Language Processing research.google. Checked 14 October 2026.
  5. Bard — Leadership & Hiring | TEXXR texxr.com. Checked 14 October 2026.
  6. Google denies Bard was trained with ChatGPT data theverge.com. Checked 14 October 2026.