The new artificial intelligence (AI) models GPT-6 Astra and Claude Fable 5.1 are demonstrating extraordinary capabilities, but they bring with them a concerning mystery. Their own creators admit to having increasing difficulty understanding the internal reasoning of the systems they developed.
The researchers from the companies responsible for these models recognize that, as neural networks become more complex and sophisticated, the decision-making process of these AIs becomes increasingly opaque, functioning almost like a "black box".
This phenomenon raises fundamental questions about the safety and control of advanced AI technologies, since the ability to explain and predict the behavior of systems is considered essential for their responsible development.
The creators of the models emphasize that, despite the impressive capabilities demonstrated, the lack of full understanding of internal functioning represents a significant challenge that the scientific community still needs to resolve.
