Vishal Sikka

Former SAP CTO, former Infosys CEO, ex-Oracle board member; studied under John McCarthy per Futurism’s profile. Founder of VianAI Systems. Co-author with his son Varin Sikka (Stanford) of sikka-2025-hallucination-stations, the July 2025 preprint whose January 2026 press wave framed it as a proof that agentic AI is mathematically doomed. The paper itself never uses the word AGI and concedes composite systems can exceed the single-model bound, so the doom framing is the coverage’s, not the authors’. Sikka told WIRED that components built around LLMs can “overcome those limitations”. Acknowledgements name John Etchemendy, Carolyn Talcott, R.V. Guha, and Alan Kay.

Sources

• arXiv:2507.07505 metadata (authors Varin Sikka, Vishal Sikka; VianAI; Stanford), fetched 2026-09 by the run in outputs/llm-degradation-toward-the-mean.provenance.md.
• Futurism, “AI Agents Are Mathematically Incapable of Doing Functional Work, Paper Finds.”