Can AI-based grading tools be used as evidence in a consumer dispute?
AI-based numismatic grading tools — software that analyses photographs to assign a grade — can be used as supporting evidence in a consumer dispute, but are unlikely to be treated as primary or determinative evidence by Indian consumer forums in 2025. The evidential value of AI grading output depends on: the tool's demonstrated accuracy and methodology; whether it is commercially deployed or experimental; and whether the forum has a basis for accepting it as reliable expert evidence. As AI grading tools mature and gain market acceptance, their evidentiary weight will increase — but currently, human expert testimony from a qualified numismatist carries more weight in Indian proceedings.
What AI grading tools currently do
Several commercial AI grading tools have emerged in numismatics internationally — PCGS's PhotoGrade system, various third-party apps, and experimental systems built on computer vision models. These tools analyse high-resolution photographs of notes or coins and output a grade estimate. The accuracy ranges from reasonable (within 2-3 grade points for common series) to unreliable (for rare or unusual items where training data is thin). No AI grading tool has yet achieved the accuracy and consistency of experienced human graders for edge cases.
The admissibility question — electronic records and expert evidence
Under the Bharatiya Sakshya Adhiniyam 2023, electronic records — including output from computer systems and software — are admissible as evidence subject to conditions of authenticity and reliability (Sections 61-65). An AI tool's grading output, if produced from a reliable, properly functioning system and authenticated, can be admitted as an electronic record. However, admissibility is not the same as persuasive weight. The forum must still decide how much weight to give the AI output compared to human expert testimony.
For AI grading evidence to carry significant weight, the party presenting it would need to: establish the tool's methodology and accuracy through a technical witness or the tool vendor's documentation; demonstrate the specific input (photograph quality, lighting conditions) was adequate for reliable output; and address any limitations the tool has with the specific note type. This is a significant evidential burden that makes AI grading currently more useful as corroborating evidence than primary evidence.
AI grading as corroboration — the practical use
The most practical current use of AI grading tools in a consumer dispute is as corroborating evidence alongside a human expert's opinion: 'The AI grading tool [name] assessed the photographs and returned a grade of VF-25; the independent expert numismatist assessed the same photographs and concluded the note is in VF condition. Both assessments are consistent with each other and inconsistent with the seller's UNC description.' This corroboration strengthens the human expert's opinion and demonstrates that the grade assessment is not a single subjective judgment.
The trajectory — AI grading will become more important
As AI grading tools improve in accuracy, gain widespread commercial deployment, and develop track records of reliability, their evidentiary weight will increase. Within a few years, a major commercial AI grading platform with documented accuracy data may carry the same evidentiary weight as a human expert witness — particularly for common series where training data is abundant. Collectors and legal professionals should stay aware of this development. For now: use AI grading as supplementary evidence; rely on human expert testimony as primary evidence; but do not dismiss AI output as irrelevant.
The law of evidence does not fear technology. It asks one question: is this reliable? When AI grading tools can answer that question with documented accuracy, case law, and institutional track records, they will become primary evidence. Until then, they are a useful corroboration tool — and the collector who uses both human expertise and AI output is building a more complete evidential record than the one who uses either alone.
Laws & authorities referenced in this chapter
Bharatiya Sakshya Adhiniyam 2023 — §§61-65 (electronic records: admissible subject to authenticity and reliability conditions)
Bharatiya Sakshya Adhiniyam 2023 — §45 (expert opinion: human numismatist opinion; AI tool output as corroborating electronic record)
Consumer Protection Act 2019 — consumer forum's broad discretion to accept any reliable evidence of grade mismatch
IT Act 2000 — §79A (electronic evidence: authenticity standards applicable to AI-generated output)
AI grading tools: admissible as electronic records under BSA 2023 §61-65. Current weight: supporting/corroborating evidence, not primary. Weight requires: establishing tool's methodology and accuracy; demonstrating adequate input quality; addressing limitations for specific note type. Best current use: corroboration alongside human expert testimony — both pointing to same grade = stronger case. Trajectory: as tools improve and gain commercial track records, their evidentiary weight will increase. For now: human expert primary; AI tool corroborating.
This is educational content, not legal advice. For a specific situation, please consult a qualified legal professional. Excerpted from Currency, Coins & The Law by Mayank Agarwal, Part 29: Grading, Authentication & Valuation — No Official Body Exists.