The Illusion of Inclusion How LLMs Misrepresent African Languages and Cultural Contexts

  • Shamsuddeen Hassan Muhammad, Imperial College London

Multilingual LLMs have become so powerful, yet African languages remain misrepresented. Despite the progress in African NLP, driven by community-led initiatives, LLMs continue to misrepresent African languages and the meanings they convey.

In this talk, Dr Shamsuddeen will discuss two decades of AfricaNLP and the collective community efforts that have shaped the field. He will then discuss ‘The Illusion of Inclusion, how LLMs misrepresent African languages and cultural contexts’ through three interrelated gaps: biased pretraining data, inaccurate evaluation, and cultural blindness. Finally, he will present recent efforts and future directions to bridge these gaps.

Watch Next