Controller

The development of a framework that can be integrated into any AI Model that will provide ethical/moral/virtue decision making, through time. Specifically focused on AI systems that are designed to be autonomous and independent where the Autonomous AI will be able to make decisions in a directional manner toward the notion of Good. Effectively being a realtime guidance system in it's daily moment by moment decisions.

User Stories

Controller for Decisons

As AI is integrated into tools and systems that allow it to be autonomous, the need to provide a controller or guidance systems will be critical. All proposed decisions the AI System will be considering can be assessed and marked against a conceptual map and a decision to execute on that action can be evaluated.

Moral / Virtue / Ethics

Clear models can be defined that can be trained into the AI Autonomous System. All decisions within reach of the AI can be evaluated and removed if the parameters are not successful met.

Domain Execution

Custom Maps can be made that allow an AI Autonomous system to meet the parameters of a decision or to abandon the decision for another one that meets the domain requirements.

Reporting and Compliance

The AI Autonomous system will be reporting their decision making and recommendations to a central command center for the human element to determine if the AI syste m is aligned or is needing additional training or adjustments over time. Trending and Warnings will be the primary purpose of this area.

Overall Product Model

Demand: As the capabilities of AI become more advanced there will be an approach to human intelligence and reasoning that will mimic the human capability, with this approach the AI systems will be capable of doing "Bad" things that can hurt others and will be able to perform tasks without constraints or compliance requirements being met. The need for this type and kind of model will be critical.

Inputs: Controller Models (Moral / Virtue / Ethical) and customized domain models.

Outputs: Prevented actions and decisions that have negative consequences to humans.

Model Testing: Internal scoring and monitoring of decision making, battery of tests to verify the models are working over time.

Compliance: Periodic tests that are reported to it's command center, out of compliance trending requires retraining and evaluation.

Market Value: The risks to humans are real where the more autonomous an AI becomes the more risks the human population will incur as the AI System will not know or "feel" that is doing wrong, that is doing "Bad". Hence the programmatic requirements for the AI system to know and move it's actions and decisions toward the place of "Good" is not a requirement but an imperative architectural need.