- Architectures for AI ... Generative AI Stack ... Enterprise Architecture (EA) ... Enterprise Portfolio Management (EPM) ... Architecture and Interior Design
- In-Context Learning (ICL) ... Context ... Causation vs. Correlation ... Autocorrelation ... Out-of-Distribution (OOD) Generalization ... Transfer Learning.
- Zero Trust
- Decentralized: Federated & Distributed ... Learning
- Service Capabilities
- Neural Architecture
- Containers; Docker, Kubernetes & Microservices
- Symbiotic Intelligence ... Bio-inspired Computing ... Neuroscience ... Connecting Brains ... Nanobots ... Molecular ... Neuromorphic ... Evolutionary/Genetic
- Google DeepMind AlphaGo Zero
- Memory Networks
- Hierarchical Temporal Memory (HTM)
- Conditional Adversarial Architecture (CAA)
- Time ... PNT ... GPS ... Retrocausality ... Delayed Choice Quantum Eraser ... Quantum
- Deep Distributed Q Network Partial Observability
- Graph Convolutional Network (GCN), Graph Neural Networks (Graph Nets), Geometric Deep Learning
- Capsule Networks (CapNets)
- Messaging & Routing
- Processing Units - CPU, GPU, APU, TPU, VPU, FPGA, QPU
- DeepMind’s PathNet: A Modular Deep Learning Architecture for AGI | Carlos E. Perez
Will AGI Require New Architectures?
- Artificial General Intelligence (AGI) to Singularity ... Curious Reasoning ... Emergence ... Moonshots ... Explainable AI ... Automated Learning
It is possible that AGI will require new architectures. Current AI systems are typically designed to solve specific problems, and they are often not very good at generalizing to new tasks or domains. AGI systems, on the other hand, will need to be able to learn and adapt to a wide range of new challenges. One way to achieve this is to develop new AI architectures that are more flexible and adaptable. For example, some researchers are exploring the use of neuromorphic architectures, which are inspired by the structure and function of the human brain. Neuromorphic architectures are more efficient at processing data and learning new tasks than traditional AI architectures. Another way to develop AGI is to use multiple AI systems working together. This approach is known as collective intelligence. Collective intelligence systems can combine the strengths of different AI systems to solve complex problems that no single system could solve on its own. Whether or not AGI will require new architectures is still an open question. However, it is clear that current AI architectures are not sufficient for achieving AGI. Researchers are exploring a variety of new approaches, and it is possible that a new architecture will be developed that is essential for achieving AGI. Here are some specific examples of new architectures that are being explored for AGI:
- Neuromorphic architectures: These architectures are inspired by the structure and function of the human brain. Neuromorphic chips are designed to be more efficient at processing data and learning new tasks than traditional AI chips.
- Hierarchical architectures: These architectures organize AI systems into a hierarchy of different levels, each of which performs a specific task. Hierarchical architectures can help to improve the efficiency and scalability of AI systems.
- Hybrid architectures: These architectures combine different types of AI systems, such as neural networks and symbolic reasoners. Hybrid architectures can help to improve the performance of AI systems on a wider range of tasks.
From Dwarkesh Patel; The Lunar Society interview with Darlo Amodei, the CEO of Anthropic: