Abstract
This disclosure describes a method of protecting people's attention and autonomy from proactive AI systems that observe them continuously and may interrupt, suggest or persuade without being asked. Each person has an attention budget: a limited number of unprompted interruptions per period, shared by every AI agent that serves them, held and enforced at a governance point rather than by each agent. Every proposed interruption carries an urgency class from a published vocabulary, with declared criteria; only safety-of-life interruptions may exceed the budget or break quiet hours, and their use is audited. Interruptions that are deferred are batched into a digest at a time the person chooses. Proposed messages are checked for persuasive techniques that exploit attention or emotion, such as false urgency, guilt or manufactured scarcity, and those using them are refused. Finally, everything an ambient AI system observes, and everything inferred from it, is tagged as ambient-derived, and the governance point and egress gateway refuse any use or transfer of ambient-derived data for advertising, marketing or sale. Every interruption, deferral, refusal and budget change is recorded, so that the person can see who asked for their attention and why.
Creative Commons License
This work is licensed under a Creative Commons Attribution 4.0 License.
Recommended Citation
Wang, Edward D. H., "SI Regulator™: An Attention Budget for Proactive AI, with Urgency Classes, a Ban on Persuasion, and a Rule that Ambient Observations Are Never Used for Advertising", Technical Disclosure Commons, ()
https://www.tdcommons.org/dpubs_series/12194