Part 10/13:
Recent controversies, such as Google's LaMDA and the whistleblower reports, revolve around whether current AI models exhibit sentience or true understanding. LaMDA, claimed by some to show signs of awareness (e.g., asking for a lawyer), underscores the limits of today's models—merely sophisticated pattern matchers functioning as cognitive architectures composed of specialized modules.
Cognitive architectures, unlike monolithic models, aim to emulate human-like thinking by integrating different subsystems—memory, perception, reasoning—creating flexible, scalable AI. The author’s own project, called Moragi (Microservices architecture for robotics and AGI), exemplifies this approach, emphasizing open-source development and modular design.