Using Information Theory to Understand the Processing Limitations of Cognitive Agents

This paper presents a theoretical exploration of the limitations of individual cognitive agents (humans or eventually AGI) in the face of collective intelligence, utilizing principles from Information Theory (IT). Through an evaluation of the relationship between signal information rate (R) and channel capacity (B), the study provides an argument based on Shannon's theorem indicating that signals with information rates vastly exceeding the channel's capacity cannot be reliably transmitted. This analogy is extended to human cognition and potentially to AGI, proposing that problem definitions and solutions existing at potentially exponentially higher resolutions in the collective conceptual space, as assumed might be the case if deduced by a hypothetical General Collective Intelligence (GCI) platform capable of organizing N cognitive agents into a network with exponentially greater intelligence than any individual cognitive agent (any human and eventually any AGI), cannot reliably be discerned by individual cognitive agents. The paper further discusses the implications of this result, suggesting that a GCI platform might offer a more reliable method for defining complex problems and discovering their solutions, where both might be beyond the reach of individual cognitive capacities. Though, the underlying models of both individual and collective conceptual spaces are still under development and have not been empirically validated, this is still potentially significant due to its implications on problems that might be above the cognitive complexity manageable by individual cognitive agents, such as keeping a super-intelligent AI safe and aligned with collective human well-being. The research concludes by emphasizing the potential significance of GCI platforms in addressing other challenges of monumental scale potentially requiring a similar network based approach to adaptive problem-solving, while highlighting the need for further research in conceptual space modeling and its applications in AI and AGI systems.