Part 10/11:
Shapiro expresses disappointment in OpenAI’s apparent reluctance to clearly define AGI or demonstrate transparency, contrasting their efforts with other organizations like Anthropic that pursue more open research. He emphasizes that true safety and progress require more than just scale or reinforcement signals—they demand advances in cognitive architectures and deeper understanding.
Despite frustrations, he admits to personally using OpenAI's models daily, appreciating their powerful capabilities but cautioning that we should not overtrust static language models as the foundation of future intelligent systems.