Execution eligibility gate for relational autonomous computational entities
Described herein are systems, methods, and media for gating execution of actions by autonomous or semi-autonomous computational entities. A proposed action may be evaluated by an execution eligibility gate configured to compute whether execution is permitted based on relational context associated with an initiating entity. Execution of the action may then be enabled, modified, or prevented by controlling an execution channel based on the computation. The architecture supports centralized, distributed, and federated environments and provides auditability of execution enforcement and execution outcomes.
1 . A computer-implemented method for action execution gating executed at one or more computer processors, the method comprising:
(a) receiving a proposed action generated by a computational entity, wherein the computational entity is an autonomous system configured to generate the proposed action through an internal process, and wherein execution of the proposed action would cause an externally observable state transition in a system distinct from the computational entity;
(b) receiving a relational context associated with the computational entity, the relational context comprising one or more dynamically updated relationships between the computational entity and one or more additional computational entities, and wherein the dynamically updated relationships are updated based on interactions between the computational entity and the one or more additional computational entities; and
(c) gating execution of the proposed action by at least:
i) computing an execution eligibility based at least in part on the relational context, wherein the execution eligibility represents a determination of whether the proposed action is permitted to be executed, and wherein computing the execution eligibility is performed independently of the internal process by which the proposed action was generated; and
ii) controlling, based on the execution eligibility, an execution channel through which the proposed action would otherwise be executed, wherein the controlling is performed in-line with the execution channel and as a condition precedent to execution of the proposed action, and wherein controlling the execution channel comprises one or more of: permitting, modifying, and preventing execution of the proposed action in a manner that constrains or alters the externally observable state transition.
2 . The method of claim 1 , wherein the proposed action comprises an operation, instruction, transaction, communication, or representation pertaining to at least one of the one or more additional computational entities.
3 . The method of claim 1 , wherein the one or more dynamically updated relationships comprise one or more of the following attributes: delegation, representation, coordination, lineage, temporal scope, constraints, revocation, conditional authority, and multi-principal authority.
4 . The method of claim 3 , wherein the relational context is from a Composite Relational Entity (CRE), wherein the CRE represents the one or more dynamically updated relationships between the computational entity and one or more additional computational entities.
5 . The method of claim 1 , wherein computing the execution eligibility is not based solely on a static authorization mechanism.
6 . The method of claim 1 , wherein controlling the execution channel comprises modifying execution of the proposed action based on the execution eligibility.
7 . The method of claim 1 , further comprising directly performing the proposed action.
8 . The method of claim 1 , wherein the relational context comprises delegated authority from another computational entity.
9 . The method of claim 1 , wherein the proposed action is generated using an internal reasoning or planning process of the computational entity and wherein computing the execution eligibility is independent of the internal reasoning or planning process.
10 . The method of claim 1 , wherein computing the execution eligibility is done jointly across a plurality of computational entities, and wherein computing the execution eligibility is based on each relational context associated with each computational entity of the plurality of computational entities.
11 . The method of claim 10 , wherein computing the execution eligibility further comprises resolving conflicting relational context associated with each computational entity of the plurality of computational entities.
12 . The method of claim 1 , wherein the execution eligibility is subject to a temporal expiration or to one or more contextual conditions.
13 . The method of claim 1 , wherein the relational context comprises lineage information associated with the computational entity.
14 . The method of claim 1 , further comprising updating the relational context based on one or more interactions between the computational entity and the one or more additional computational entities.
15 . The method of claim 1 , wherein controlling the execution channel comprises preventing execution of the proposed action, and wherein preventing execution of the proposed action comprises preventing a commitment of a state transition in the system distinct from the computational entity.
16 . The method of claim 1 , wherein computing the execution eligibility is based at least in part on an approval from an individual.
17 . The method of claim 1 , further comprising recording the execution eligibility.
18 . The method of claim 1 , further comprising generating an explanation of a basis for performing the executive action.
19 . A computer-implemented system for action execution gating comprising at least one computer processor and instructions executable by the at least one computer processor to provide an application comprising:
(a) a relational context data store configured to store data representing relationships among computational entities, wherein the data is updated dynamically over time; and
(b) an eligibility gating engine configured to perform operations comprising:
i) receiving a proposed action initiated by a computational entity, wherein the computational entity is an autonomous system configured to generate the proposed action through an internal process;
ii) receiving a relational context associated with the computational entity from the relational context data store, the relational context comprising one or more dynamically updated relationships between the computational entity and the one or more additional computational entities, and wherein the dynamically updated relationships are updated based on interactions between the computational entity and the one or more additional computational entities;
iii) gating execution of the proposed action by:
1) Computing an execution eligibility based at least in part on the relational context wherein the execution eligibility represents a determination of whether the proposed action is permitted to be executed, and wherein computing the execution eligibility is performed independently of the internal process by which the proposed action was generated; and
2) Controlling, based on the execution eligibility, an execution channel through which the proposed action would otherwise be executed, wherein the controlling is performed in-line with the execution channel and as a condition precedent to execution of the proposed action, and wherein controlling the execution channel comprises one or more of: permitting, modifying, and preventing execution of the proposed action in a manner that constrains or alters the externally observable state transition;
wherein the eligibility gating engine is configured to operate in-line with the execution channel prior to execution of the proposed action.
20 . The system of claim 19 , wherein the proposed action comprises an operation, instruction, transaction, communication, or representation pertaining to at least one of the one or more additional computational entities.
21 . The system of claim 19 , wherein the one or more dynamically updated relationships comprise one or more of the following attributes: delegation, representation, coordination, lineage, temporal scope, constraints, revocation, conditional authority, and multi-principal authority.
22 . The system of claim 21 , wherein the relational context is from a Composite Relational Entity (CRE), wherein the CRE represents the one or more dynamically updated relationships between the computational entity and one or more additional computational entities.
23 . The system of claim 19 , wherein controlling the execution channel comprises modifying execution of the proposed action based on the execution eligibility.
24 . The system of claim 19 , wherein the eligibility gating engine is further configured to directly perform the proposed action, and wherein the eligibility gating engine is not configured to compute the execution eligibility based solely on a static authorization mechanism.
25 . The system of claim 19 , wherein the relational context comprises delegated authority from another computational entity.
26 . The system of claim 19 , further comprising a role-based access control system, wherein the eligibility gating engine is configured to operate independently of the role-based access control system.
27 . The system of claim 19 , further comprising a centralized policy engine, wherein the eligibility gating engine is configured to operate independently of the centralized policy engine.
28 . The system of claim 19 , wherein the computational entity is configured to regenerate, and wherein the eligibility gating engine is configured to calculate the execution eligibility independent of regeneration of the computational entity.
29 . The system of claim 28 , further comprising a plurality of execution channels, wherein the eligibility gating engine is configured to gate execution of the action at one or more of the plurality of execution channels.
30 . A non-transitory computer-readable medium storing instructions that, when executed by one or more computer processors, cause the one or more computer processors to perform operations comprising:
(a) receiving a proposed action generated by a computational entity, wherein the computational entity is an autonomous system configured to generate the proposed action through an internal process, and wherein execution of the proposed action would cause an externally observable state transition in a system distinct from the computational entity;
(b) receiving a relational context associated with the computational entity, the relational context comprising one or more dynamically updated relationships between the computational entity and one or more additional computational entities, and wherein the dynamically updated relationships are updated based on interactions between the computational entity and the one or more additional computational entities; and
(c) gating execution of the proposed action by at least:
i) computing an execution eligibility based at least in part on the relational context, wherein the execution eligibility represents a determination of whether the proposed action is permitted to be executed, and wherein computing the execution eligibility is performed independently of the internal process by which the proposed action was generated; and
ii) controlling, based on the execution eligibility, an execution channel through which the proposed action would otherwise be executed, wherein the controlling is performed in-line with the execution channel and as a condition precedent to execution of the proposed action, and wherein controlling the execution channel comprises one or more of: permitting, modifying, and preventing execution of the proposed action in a manner that constrains or alters the externally observable state transition.