Part 9/14:
He explores the concept that current models like GPT-3 are stateless and lack memory, which limits their ability to develop true awareness. Instead, he envisions integrating search, indexing, and persistent memory into AI systems—allowing them to recall past interactions, context, and knowledge over time. This leads him towards the idea of "artificial awareness," where a system recognizes its own ongoing activity and attention, edging closer to human-like understanding.
Cord supports the notion that machines already possess primitive forms of awareness, evidenced by experiments where robots observe their own movements or AI models monitor their own processes. The challenge lies in formalizing this awareness and embedding it into scalable, reliable systems.