AGI Architecture Research Articles

From Pursuit of the Universal AGI Architecture to Systematic Approach to Heterogeneous AGI (SAGI): Addressing Alignment, Energy & AGI Grand Challenges

Artificial intelligence (AI) faces a trifecta of grand challenges: the Energy Wall, the Alignment Problem and the Leap from Narrow AI to AGI. Contemporary AI solutions consume unsustainable amounts of energy during model training and daily operations. Making things worse, the amount of computation required to train each new AI model has been doubling every 2 months since 2020, directly translating to unprecedented increases in energy consumption. The leap from AI to AGI requires multiple functional subsystems operating in a balanced manner, which requires a system architecture. However, the current approach to AI lacks system design; even though system characteristics play a key role in the human brain; from the way it processes information to how it makes decisions. In this paper, we posit that system design is the missing piece in overcoming current AI the grand challenges. We present a Systematic Approach to AGI (SAGI) that utilizes system design principles to overcome the energy wall and the alignment challenges. This paper asserts that artificial intelligence can be realized through a multiplicity of design-specific pathways, rather than a singular, overarching AGI architecture. AGI systems may exhibit diverse architectural configurations and capabilities, contingent upon their intended use cases. We argue that AI alignment, the most difficult among the grand challenges, is not attainable without a way to reflect the complexity of the human moral system and its subsystems in the AGI architectures. We claim that AGI approaches such as symbolicism, connectionism and others are not fundamental to AGI but emergent from the system design processes. Hence, we focus on employing system design principles as a guiding framework, rather than solely concentrating on a universal AGI architecture.

From the Pursuit of Universal AGI Architecture to Systematic Approach to Heterogenous AGI: Addressing Alignment, Energy, & AGI Grand Challenges

From the Pursuit of Universal AGI Architecture to Systematic Approach to Heterogenous AGI: Addressing Alignment, Energy, & AGI Grand Challenges

Shifting and drifting attention while reading: A case study of nonlinear-dynamical attention allocation in the OpenCog cognitive architecture

Abstract A simple experimental example of the general principle of “cognitive synergy” underlying the OpenCog AGI architecture is explored: An OpenCog system processing a series of articles that shifts from one topic (insects) to another (poisons), and using its nonlinear attention-allocation dynamics (based on the ECAN Economic Attention Networks framework) to spread attention back and forth between the nodes and links within OpenCog’s Atomspace knowledge store representing the words in the sentences, and other nodes and links containing related knowledge. With this setup, we study how the ECAN system shifts the attentional focus of the system based on changes in topic – in terms of both the speed of attention switching, and the contextual similarity of the content of attentional focus to the sentences being processed at a given point in time. This also provides an avenue for exploring the effects of particular design choices within the ECAN system. For instance, we find that in this particular example, if the parameters are set appropriately, ECAN indeed causes the system to assign particular importance to nodes and links related to the “insecticide” concept, when it is reading sentences about poisons in a situation where it has been primed by sentences about insects. This is an example of what we call “drifting” attention – the system’s attention moves to something suggested by its perceptions, even if not directly presented in them.

Sheet-like and truncated-dodecahedron-like AgI structures via a surfactant-assisted protocol and their morphology-dependent photocatalytic performance.

Silver halide-based structures have been attracting great interest as efficient visible-light-driven photocatalysts towards the photodegradation of organic pollutants, and those studies focusing on their morphology-dependent catalytic performances have received particular attention. While great advancements in this regard have been witnessed in the past few years with respect to AgCl- and AgBr-based photocatalysts, relevant explorations concerning AgI-based species are relatively rare, even though the excellent durability of AgI-based structures renders them attractive candidates for potential photocatalytic uses. By means of chemical reactions between AgNO3 and tetramethylammonium iodide (TMAI), and AgNO3 and tetrabutylammonium iodide (TBAI), we herein report that AgI structures with a sheet-like and a truncated-dodecahedron-like morphology, respectively, could be controllably synthesized via a surfactant-assisted fabrication protocol. In our synthesis systems, AgNO3 works as the silver source, while the TMAI and TBAI surfactants serve not only as an iodine source but also as a directing reagent for controllable fabrication. It has been demonstrated that our AgI structures could work as visible-light-energized photocatalysts towards the photodegradation of methyl orange. We find that compared to their sheet-like counterparts, the truncated-dodecahedron-like AgI architectures exhibit substantially boosted catalytic performances. Moreover, we disclose that our truncated-dodecahedron-like AgI-based species could display excellent photocatalytic stability, wherein their catalytic reactivity displays only trivial fluctuations under visible-light irradiation even after the photoreactions have been repeated 22 times continuously. Our work might not only introduce a facile protocol for the controllable synthesis of AgI structures but also pave an avenue for facile enhancement of their catalytic performances via morphology alterations.

Silver Iodide Microstructures of a Uniform Towerlike Shape: Morphology Purification via a Chemical Dissolution, Simultaneously Boosted Catalytic Durability, and Enhanced Catalytic Performances

The fabrication of microstructures/nanostructures of a uniform yet well-defined morphology has attracted broad interest from a variety of fields of advanced functional materials, especially catalysts. Most of the conventional methods generally suffer from harsh synthesis conditions, requirement of bulky apparatus, or incapability of scalable production, etc. To meet these formidable challenges, it is strongly desired to develop a facile, cost-effective, scalable method to fulfill a morphology purification. By a precipitation reaction between AgNO3 and KI, we report that irregular AgI structures, or their mixture with towerlike AgI architectures could be fabricated. Compared to the former, the mixed structures exhibit enhanced catalytic reactivity toward the photodegradation of Methyl Orange pollutant. However, its catalytic durability, which is one of the most crucial criteria that are required by superior catalysts, is poor. We further show that the irregular structures could be facilely removed from the mixture via a KI-assisted chemical dissolution, producing AgI of a uniform towerlike morphology. Excitingly, after such simple morphology purification, our towerlike AgI displays not only a boosted catalytic durability but also an enhanced catalytic reactivity. Our chemical dissolution-based morphology purification protocol might be extended to other systems, wherein high-quality advanced functional materials of desired properties might be developed.

Cognitive Synergy between Procedural and Declarative Learning in the Control of Animated and Robotic Agents Using the OpenCogPrime AGI Architecture

The hypothesis is presented that "cognitive synergy" -- proactive and mutually-assistive feedback between different cognitive processes associated with different types of memory -- may serve as a foundation for advanced artificial general intelligence. A specific AI architecture founded on this idea, OpenCogPrime, is described, in the context of its application to control virtual agents and robots. The manifestations of cognitive synergy in OpenCogPrime's procedural and declarative learning algorithms are discussed in some detail.