Keynote – Arthur Mensch
Summary
Mensch argued for AI sovereignty through open‑source. He cited Mistral’s commitment to releasing state‑of‑the‑art models under permissive licenses, enabling countries to own and deploy AI locally.
Key points:
- Open‑source as a trust enabler: Transparency, community auditability, and reduced vendor lock‑in.
- Multilingual focus: Mistral models prioritize low‑resource languages, crucial for India’s 22 official languages.
- Decentralised AI ecosystem: Governments should host their own models, akin to the early internet’s open architecture.
Mensch warned that concentration of AI power threatens democratic governance and called for global collaboration on standards for safety and fairness.
Key Takeaways
- Open‑source AI is essential for national sovereignty and trust.
- Multilingual capabilities prioritize Indian and other low‑resource languages.
- Decentralisation: Countries should host their own models, reducing dependence on foreign providers.
- Regulatory cooperation is needed to ensure safety and fairness across jurisdictions.