ref.apple.hig-machine-learning · official-system
Apple HIG — Machine Learning
Apple Human Interface Guidelines page for ML-enabled experiences covering product-level considerations such as appropriate confidence thresholds, graceful degradation, privacy, correction and feedback mechanisms, and when not to use ML. The guidance is version-sensitive and scoped to Apple platforms.
https://developer.apple.com/design/human-interface-guidelines/machine-learning
What this source teaches
- Cross-check existing capability, agency, uncertainty, privacy, and feedback contracts against Apple's ML design considerations — in particular: show only high-confidence results, provide easy correction, and design graceful degradation.
- Do not use the existence of an ML capability as justification for a feature; require a clear user benefit and a tested degraded-state.
- Add privacy considerations for on-device vs. server-side ML as a capability disclosure requirement.
Where AwesomeDS applied it
rule.ai.ux-agency-contract— Preserve proportionate user agency over AI actionrule.ai.ux-capability-contract— Declare AI capability and boundaries before reliancerule.ai.ux-failure-contract— Model AI failure by cause, consequence and recoveryrule.ai.ux-uncertainty-contract— Communicate evidence and uncertainty at the decision point
Implementations and artifacts
No directly linked artifacts.
Caveats & anti-imitation
Transfer product-level ML design considerations (capability, privacy, correction, degradation); do not import Apple platform APIs or iOS-specific interaction patterns as cross-platform requirements.