In the rapidly evolving artificial intelligence landscape, Bren Worth, chief technology officer and co-founder of Substrate AI and SubGen AI. With a career dedicated to pioneering advances in AI, Worth has focused on ethical AI development, solving complex problems and creating innovative applications in a variety of fields. His recent work on projects such as CoAutoGenius and Serenity Star AI, along with a strong portfolio of granted and pending patents, underscores a firm commitment to pushing the boundaries of AI technology and its practical applications.

Innovations in Substrate AI: a blend of ethics and technology

Since co-founding Substrate AI in Spain in 2016, Worth has been instrumental in leading the company towards developing innovative solutions for various verticals, including Agritech, AgTech, Human Resources, Continuous Employee Learning, Financial Trading and Recruiting. His role as lead architect on the transformative CoAutoGPT project exemplifies his ability to integrate advanced tools for code writing, deployment and product creation with a human collaborative agent architecture. This project, which emphasizes the ethical development of AI and biologically inspired working memory systems, represents a significant breakthrough in complex project execution.

Another hallmark of Worth's tenure at Substrate AI is the development of Serenity Star AI agents, where he spearheaded the design and implementation of an innovative cognitive-inspired LLM agent. This agent's capabilities in multimodal reasoning, ethical behavior moderation, and adaptive continuous learning integrate cutting-edge ideas. Worth's achievements in the development of a risk assessment framework, cross-modal value learning and an ACA (Assessment, Coping and Adaptation) content assessment algorithm will provide a material advance in the direction of powerful but safe AI.

SubGen AI: a new adventure in generative AI

With SubGen AI, Substrate AI ventures into the realm of generative AI technology, focusing on creating, deploying and managing generative AI-based SAAS products for enterprise and government customers. The Serentity Star product line, which includes customized on-premises LLMs, AI Hub for managing and deploying generative AI services with its advanced Meta-Control agent and Instructor, and the Serenity* advanced multimodal reasoning agent, underscore SubGen AI's commitment to secure and cost-effective AI solutions.

Bren Worth's role at Substrate AI, particularly in the development of SubGen AI's Serenity Star and other AI products, positions him as a pivotal figure in shaping the trajectory of ethical and advanced AI systems. His contributions reflect a deep commitment to integrating cutting-edge technology with ethical principles, ensuring that AI applications are innovative and responsible. Let's delve into his role and the impact of his work on Serenity Star AI, Meta-control and other initiatives under SubGen AI.

A deeper dive into technology

Serenity Star AI: Pioneering agents inspired by cognition
Serenity Star AI represents a concrete step forward in AI development, incorporating a cognitively inspired approach to Large Language Models (LLM). Our project combines human-based control strategies and multimodal reasoning and a novel temporal and spatial abstraction memory system, meaning that it can process and integrate information from diverse sources and modalities (e.g., text, images, data) to make decisions or generate answers. This ability is crucial for complex problem solving in real-world scenarios where information comes in a variety of formats and requires abstract, multi-step reasoning.

The Serenity Star agent is designed to navigate ethical dilemmas and moderate its behavior accordingly to solve complex problems. The project has developed advanced algorithms for risk assessment, cross-modal value learning, and content appropriateness. These components are crucial to ensure that AI actions and suggestions align with human values and ethical norms.

In addition, the Serenity Star AI agent includes the development of an advanced memory system that combines temporal and spatial abstraction, enabling the AI to understand and reason about events over time and space. This feature, combined with automatic personally identifiable information (PII) and private data replacement, ensures that the AI respects user privacy and complies with data protection regulations. Leveraging the advanced memory system, the Serenity Star agent can perform multi-step reasoning and abstractly capture and manipulate the world, giving it unprecedented capabilities for planning and reasoning.

MetacontrolProtection against the risks of AI
The Meta-control system is another high-value technology for SubGen AI. This service operates at the core of the AI Hub, monitoring the proper use and safe application of LLM-based products across the enterprise. Meta-control can be likened to an adaptive antivirus for AI technologies, where threats include not only security risks, but also misuse and exploitation of the vast access to LLM data.

Worth's focus on meta-control underscores its commitment to mitigating significant risks associated with LLM technologies, such as hallucinations, data leakage and improper behavior. By enforcing policy and regulatory compliance, Meta-control helps ensure that LLMs operate within ethical boundaries, which enhances trust and reliability in AI applications.

Agent trainer: Improving AI system integrity
The Instructor agent, an innovative technology under Worth's leadership, automates testing of LLM systems to ensure their security and reliability. This agent creates virtual personas that simulate external users attempting to exploit system vulnerabilities. Through this process, the instructor agent evaluates the effectiveness of meta-control in defending against insecure or unethical use, providing a quality score that builds confidence in the integrity of AI systems.

This approach not only strengthens the safety and ethical operation of AI applications, but also generates synthetic data that can be used to further train and refine AI models, improving their performance and reliability.

Much more work to be done!
Bren Worth's role in the development of Serenity Star AI, Meta-control and the instructor agent at SubGen AI illustrates a visionary approach to AI development. By prioritizing ethical considerations, multimodal reasoning capabilities and robust safety mechanisms, Worth is not only advancing the technological capabilities of AI systems, but also ensuring that they operate in a responsible, ethical manner aligned with human values. His work sets a benchmark for the future of AI development, where innovation goes hand-in-hand with ethical responsibility and user safety.