The Indian National Science Academy convened an online BRICS science-academies meeting on June 8, 2026 to develop cooperation on artificial intelligence. The Science Ministry's account described agreement around shared computing, data platforms and multilingual resources, alongside interest in smaller models designed for particular industrial tasks where errors can have serious consequences.
Individual contributions made those priorities more concrete. China proposed evaluation arrangements addressing reliability, bias and local conditions; Indonesia emphasised early warning and disaster risk; Vietnam called for joint training and shared infrastructure. These proposals connect the ability to build a model with the separate ability to judge whether it is useful in the place where it will operate.
Shared computing addresses access to processing capacity. Evaluation concerns the evidence used to assess the resulting system. The distinction explains why the forum combined infrastructure proposals with standards and skills: access to a powerful machine cannot, by itself, answer the local questions raised by the participating academies.
The joint statement subsequently published by INSA develops a specific disaster-management pathway. It describes combining satellite, sensor and socioeconomic information for warnings, organising vulnerability and exposure data, and using language technology to issue multilingual alerts. It also identifies post-disaster tasks such as locating access routes and routing supplies.
That sequence reaches beyond predicting a hazard. Information about who is exposed can guide a warning, while information about accessible routes supports the response after an event. The statement makes validated historical data a condition of performance and retains a role for locally informed civil-defence systems.
Its agriculture section makes a related point about delivery: AI services need connectivity, digital literacy and effective extension systems, especially for smallholders. A possible BRICS AI Research and Innovation Network would support benchmarks, pilot projects, skills and data frameworks. These are proposed cooperation mechanisms, with national rules and data privacy retained, rather than a single model expected to serve every local setting.
