Last updated on December 5th, 2023 at 12:51 pm
Reframing AGI: A Functional and Systemic Approach
In the landscape of artificial intelligence, where the pursuit of Artificial General Intelligence (AGI) dominates discussions, Google’s five levels of AGI have recently been proposed by the folks at Google Deepmind. It has merit, but I propose a paradigm shift. Moving away from the generative AI model focused on individual achievement, I advocate for a framework that recognizes AGI as a functional and systemic entity. This is a call for a more pragmatic and societal-centric approach to AGI development.
Beyond Creative Mastery: AGI as Societal Infrastructure
Google’s five levels of AGI are focused more on achievement at a parallel to human achievement, a comparison I don’t think is useful for an intelligence that is systemic. (They are useful for generative AI, possibly, although I don’t understand why we are focusing energy in getting AI to do things we are very good at vs things like systemic data analysis and systems optimization which we are not.) And AI will be. Is already.
The prevalent models often celebrate AI that mimics human creativity and skills, coining terms like ‘virtuoso’ to denote machines that can play chess or compose symphonies. Yet, this perspective on AI as generative AI limits our vision. It’s not the mimicry of human creativity that will define AGI, but rather the ability to function autonomously and systemically within the fabric of society.
The Functional Levels of AGI
In the functional framework I propose, AGI is not a solitary genius but a collaborative force. From the initial “Coordinator AI” that manages specific tasks to the more advanced “Society Recoding AI” and ultimately, the “Human Future Redesign AI,” each level is defined by clear milestones. These milestones are not just about what AI can create but how it can support, enhance, and integrate with human society.
This framework serves as a blueprint for AGI integration into societal systems. It’s about embedding AI into our healthcare, logistics, governance, and environmental management systems, ensuring that AI development is aligned with human needs and societal progress.
Level 0 (today) Generative AI Based
We feed it, we talk to it, it talks back to us with information, images constructed from billions of pieces of content with billions of rules attached to them. Creative and other functional support.
Level 1: Coordinating AGI
Can manage multiple AI closed-loop functions for a specific purpose, like doing the job of coordinating delivery drones for an e-commerce company.
Level 2: OperationalAGI
Can run operational aspects of an organization, department or business, similar to AI systems managing inventory, logistics, and customer service in a large retail chain.*
Level 3: Social CaretakingAGI
Provides FT support and companionship, resembling AI used in elder care for monitoring health, providing reminders, and offering social interaction. Emotional + intellectual reading and responding. Combined with robotics.
Level 4: Society Recoding AGI
Through the connection of all live data sources, optimizes outcomes across various domains, like AI in smart grids managing energy distribution and consumption for optimal performance.
Level 5: Human Future Redesign AGI
Planetary. Demonstrates sentience and surpasses human capabilities. Has as its primary objective the preservation of the future of humanity. It can run most parts of human society—fully outsourced domestic management to automatons.
The Road Ahead
As we chart the course for AGI, it’s crucial to establish standards that guide its growth. The functional and systemic model I propose offers a structured path to achieving an AGI that is not only intelligent but also benevolent, ethical, and harmoniously integrated with the complexities of human society.
For businesses, the generative model offers a roadmap for incrementally adopting AI technologies that progressively take on more complex tasks. Conversely, the systemic model provides a vision for AI’s role in restructuring business processes and societal operations, with clear strategic planning and investment milestones.
The generative model aligns with a traditional view of intelligence as individual mastery, while the systemic model presents AGI as an integral part of society’s fabric. As the AI field progresses, it may be necessary to blend these models, recognizing the generative capabilities of AI while steering its development towards systemic integration and societal benefit.
In conclusion, the true measure of AGI should not be its ability to write novels or paint pictures but its capacity to sustain and advance the very society that brought it into existence. This is the AGI that we should aspire to create.