TLDR: Orange has announced a multi-year partnership with OpenAI and Meta to develop and fine-tune AI Large Language Models (LLMs) capable of understanding regional African languages. This initiative aims to bridge the digital divide by enabling natural language communication for customer support and sales, starting with Wolof and Pulaar in West Africa, and eventually expanding across Orange’s 18-country African footprint. The project, set to begin in the first half of 2025, emphasizes responsible AI development and will make open-source models available for non-commercial use.
French telecommunications giant Orange has unveiled a significant strategic partnership with leading AI developers OpenAI and Meta, aimed at accelerating digital inclusion across the African continent. The collaboration focuses on fine-tuning advanced Artificial Intelligence Large Language Models (LLMs) to comprehend and process regional African languages, a critical step towards enabling more natural and accessible digital interactions for millions.
The initiative, slated to commence in the first half of 2025, will initially concentrate on two prominent West African languages: Wolof, spoken by an estimated 16 million people, and Pulaar, with approximately 6 million speakers. These languages are currently underrepresented in existing generative AI models, highlighting the necessity of Orange’s targeted approach. The project will leverage OpenAI’s ‘Whisper’ speech model and Meta’s ‘Llama’ text model, specifically Llama 3.1, fine-tuning them with diverse linguistic data to enhance their understanding of these regional dialects.
Orange’s long-term vision extends to recognizing all spoken and written languages across its extensive 18-country African operational footprint. This ambitious goal underscores the company’s commitment to ensuring that the benefits of AI are accessible to all, including illiterate populations who currently face significant barriers to digital engagement.
Beyond enhancing customer support and sales interactions, Orange plans to make these fine-tuned, open-source AI models available for non-commercial use. This provision is intended to support vital public services such as health, education, and various community initiatives, fostering local innovation and mitigating the growing digital divide.
Steve Jarrett, Orange’s Chief Artificial Intelligence Officer, emphasized the strategic importance of the partnership with OpenAI, stating, “OpenAI’s models are the most popular. And so it made financial sense for us to have a direct billing relationship.” This direct access agreement also provides Orange with early access to pre-release AI models, with data processing and hosting securely managed within European data centers.
Orange is also committed to a philosophy of “Responsible AI.” The company stated in its press release that it “carefully chooses the most appropriate and simplest solution for each AI use case.” This approach involves “only using the latest LLMs where they are necessary and otherwise choosing simpler and cheaper solutions, thereby minimising the impact on the environment as well as reducing cost for the many valuable AI use cases deployed across Orange.”
Also Read:
- Bragi and OpenAI Collaborate to Revolutionize Headphone Experiences with Generative AI
- AI-RAN Alliance and Linux Foundation Forge Partnership to Accelerate AI-Native Networks and 6G Evolution
This partnership comes at a time of burgeoning investment in generative AI, with enterprise spending in the sector having increased sixfold to reach $13.8 billion in the past year. Industry analysts, including Menlo Ventures, report that 72 percent of enterprise information technology decision-makers anticipate broader adoption of generative AI in the near term, underscoring the timeliness and strategic foresight of Orange’s latest venture in Africa.
