The Evidentiary And Epistemic Breakdown Of Patentability In Autonomous Molecular Design: Reconciling Section 3(D) And The Skilled-Person Standard With Black-Box In- Silico Drug Discovery
Siddharth Shankar, LLM, Vishwakarma University
ABSTRACT
Autonomous molecular design yields candidate compounds without articulable reasons, supported by efficacy evidence that is computed rather than observed. This article asks whether India’s patentability architecture can absorb that shift: the skilled-person standard behind section 2(1)(ja) of the Patents Act, 1970, the therapeutic-efficacy barrier in section 3(d), and the procedural rules in section 104A and in section 63(4) of the Bharatiya Sakshya Adhiniyam, 2023. It surveys the doctrine from Bishwanath Prasad and Roche v. Cipla through Novartis and the Delhi High Court’s 2025–2026 decisions, and identifies the lacunae. It argues that the breakdown is real but mislocated. The skilled person does not collapse, because computation makes candidate generation cheap without making efficacy predictable; inventive contribution migrates from structure to evidence. Section 3(d) holds, but generative optimisation targets properties that it discounts. The procedural layer fails: section 104A presupposes concealed physical manufacture, and section 63(4) authenticates files, not models. The article proposes clarifications to sections 2(1)(ja), 3(d) and 6, an adverse-inference supplement to section 104A, and a short, conditional, access-safeguarded provisional exclusivity tied to auditable provenance records.
Keywords: autonomous molecular design; section 3(d); person skilled in the art; therapeutic efficacy; section 104A; Bharatiya Sakshya Adhiniyam; provisional exclusivity.
