Can is not should

A model may infer emotion from a face, rank applicants, imitate a person's voice or compose a persuasive message. A demonstration of capability does not establish validity, necessity, consent, proportionality or a legitimate role in a consequential decision.

Ask what the system changes for a person

  • What worthwhile purpose requires AI?
  • Who gains convenience, and who carries the error?
  • Could a less intrusive method work?
  • Can an affected person understand and challenge the outcome?
  • Who remains accountable when the system is wrong?

Refusal can be a mark of capability

An organization demonstrates maturity when it can stop a use that is impressive but poorly justified. Delaying deployment, narrowing scope or choosing a simpler rule can preserve trust and attention for uses with clearer value.

Governance preserves human agency

The aim is not to reject automation by instinct. It is to make purpose, evidence, limits and responsibility visible before convenience hardens into infrastructure.

Evidence & context: National Institute of Standards and Technology · OECD.AI · IndiaAI

Sources & further reading

  1. Artificial Intelligence Risk Management Framework (AI RMF 1.0)

    National Institute of Standards and Technology. Voluntary, rights-preserving guidance organized around GOVERN, MAP, MEASURE and MANAGE. NIST was revising AI RMF 1.0 when checked on 28 September 2026, so organizations should verify the current version before formal adoption.

  2. OECD AI Principles

    OECD.AI. Intergovernmental principles updated in May 2024 covering inclusive benefit, human rights and fairness, transparency, robustness and accountability. They are high-level guidance rather than a complete operational control set.

  3. Responsible AI: Principles and Point of Focus

    IndiaAI. India-focused public guidance on safety, equality, privacy, transparency, accountability and human oversight. It informs educational governance practice but is not a substitute for current sector-specific law or qualified legal advice.

Examples and exercises are illustrative unless attributed to a source. No independent expert review is claimed.

A correction, a counterexample or an experience worth sharing?

Join the conversation ↗