The Seven Deaths: A Falsification Framework for "Learns from Experience" Claim

Key Points

• The aim is to critically assess the assertion that artificial agents learn from experience by proposing a structured framework of null hypotheses.
• Formalization of seven distinct null hypotheses related to artificial learning capabilities.
• Development of a falsification framework to evaluate the learning claims of artificial agents.
• Identified seven null hypotheses that can indicate when an artificial agent is not genuinely learning from experience.
• Provided a structured approach to evaluating learning claims, enhancing the ability to discern genuine learning from behavioral adaptations.