Interview with Google DeepMind Fellow David Ifeoluwa Adelani

Dr David Adelani, UCL’s first Google DeepMind Fellow, discusses the transformative impact of his research in Natural Language Processing (NLP). As he embarks on a new chapter at McGill University, David reflects on his journey, accomplishments, and Google DeepMind’s support.

1. Can you give us an overview of the research project you worked on during your fellowship at UCL?

My research project is on NLP for under-resourced languages. My approach follows the participatory research method, where the native speakers of these languages are involved in building datasets and models for these languages.

One of the communities of native speakers I have collaborated closely with is Masakhane – a grassroots organisation focusing on NLP for African languages.

Some of my notable achievements include benchmark datasets for African languages, such as MasakhaPOS, NollySenti, ÌròyìnSpeech, and SIB-200, as well as a novel cross-lingual question answering dataset, AfriQA.

Ten of my papers based on this work have been accepted at top NLP conferences, which showcase some significant advances. I’m really proud of that.

Additionally, I collaborated on projects like BLOOM+1 and AfriTeVa V2, expanding multilingual language models and developing new methods for language model adaptation.

2. What are the real-world implications of your research?

This research helped to address the under-representation of many languages in language technology e.g. searching for a query on search engines (AfriQA) or analysing sentiment of a movie (NollySenti) or social media like Twitter in an African language (AfriSenti).

This research aligns with the UN Sustainable Development Goal (SDG) of reduced inequalities.

3. What have been the highlights of your time at UCL?

Some highlights include:

• Receiving the Area Chair award for our paper on MasakhaNEWS at IJCNLP-AACL (International Joint Conference on Natural Language Processing and the Asia-Pacific Chapter of the Association for Computational Linguistics) 2023 in Bali.
• First NeurIPS (Neural Information Processing Systems) paper with colleagues at UCL-NLP, the largest AI conference in our field.
• Co-teaching Statistical Natural Language Processing, the largest class I have taught.
• Multiple conference papers acceptance:
  • three ACL (Association for Computational Linguistics)
  • four EMNLP (Empirical Methods in Natural Language Processing) papers
  • two LREC-COLING (Joint International Conference on Computational Linguistics, Language Resources and Evaluation)
  • one EACL (European Chapter of the Association for Computational Linguistics)
  • and one NAACL (North American Chapter of the Association for Computational Linguistics).

  This is the first time I will have more than two papers consecutively at the top two NLP conferences in ranking.

  4. Was there anything in UCL support that you particularly welcomed or might have contributed to your success?

  I was invited to join the UCL-NLP group, which helped me collaborate on research projects with several NLP researchers, that led to academic publications. It also provided me with the opportunity to supervise and collaborate with students, from BSc, MSc, and PhD.

  The support from my mentors like Pontus Stenetorp and Marc Deisenroth when I was applying for faculty positions was also really appreciated.

  5. Could you share how the collaboration with Google DeepMind influenced your research fellowship?

  Support from Google DeepMind was invaluable as they provided us with mentors who gave us useful career advice. My mentor Sebastian Riedel was very helpful in offering a lot of feedback when I was preparing for my faculty job talks.

  They also offered some meetings to learn about our progress and the challenges we have faced, and even the chance to contribute to developing the next generation of AI talent through student talks and panels.

  6. What’s next for you?

  I secured a tenure-track assistant professor position at McGill University, Canada, and I will also be a core member at MILA - Quebec AI Institute, the world's largest academic research centre for deep learning. I believe that the mentorship and the support I received from UCL and Google DeepMind has been instrumental in me getting this position.

  Applications for the next UCL Google DeepMind Fellowship

  Applications are expected to open for the next UCL Google DeepMind Fellowship position in September. Register for alerts on the UCL Jobs website to stay informed about this exciting opportunity.