By Qaiser Nawab | AzerNEWS | Guest columnist China's proposal at the 2026 BRICS Summit in New Delhi to establish a BRICS AI Open-Source Zone offers one possible direction. The initiative, presented by President Xi Jinping, includes cooperation in large language model development and applications, AI training and the creation of an open ecosystem. China also proposed cooperation on a digital industry cloud platform, smart factories and an engineer training alliance, alongside efforts to build greater international consensus on AI governance.
The significance of open-source artificial intelligence lies in the possibility of reducing dependence on a limited number of proprietary platforms. Commercial AI services can be useful, but their pricing, technical restrictions, data policies and availability may limit how institutions in developing countries use them. Open-weight models, when their parameters are made available under appropriate licences, can provide researchers and organisations with greater opportunities to adapt technology to local needs.
They may support applications in languages that receive less attention from major commercial developers, improve local research capacity and enable institutions to experiment without relying exclusively on external service providers. The expansion of Chinese models, including Qwen, DeepSeek, Kimi and GLM, has contributed to a more competitive international AI landscape. These systems differ in their technical capabilities, licensing arrangements and commercial models, and they should be assessed individually rather than treated as a single category.
Nevertheless, the broader development of open-weight AI has created new possibilities for institutions outside the traditional centres of technological power. For the Global South, the value of these technologies will not be measured only by benchmark results. It will also depend on whether universities can use them for research, whether businesses can deploy them at manageable costs and whether public institutions can develop reliable applications for education, healthcare, agriculture and public services.
Consider a university in Pakistan, Kenya or Brazil seeking to build an AI tool for local-language education. Access to an adaptable model could reduce some initial development barriers. But the university would still need computing resources, quality training data, technical expertise and safeguards against inaccurate or discriminatory outputs.
Open-source availability is therefore a starting point, not a complete development strategy. BRICS countries possess different levels of industrial capacity, research expertise, digital infrastructure and regulatory experience. The proposed AI training programmes and engineer alliance are potentially important in this regard.
Extract — continue reading at the source.
