The field of artificial intelligence is undergoing a significant shift in its understanding of intelligence and cognition. Recent research has highlighted the limitations of current neural network paradigms and the need for new frameworks that recognize the non-human nature of artificial intelligence. The concept of intelligence is being reexamined, and new methods for evaluating AI systems are being developed. Notably, the idea of applying human psychometric frameworks to AI evaluation is being challenged, and alternative approaches that focus on native machine cognition assessments are being proposed.

Some noteworthy papers in this area include: The Catastrophic Paradox of Human Cognitive Frameworks in Large Language Model Evaluation, which identifies a paradox in the evaluation of large language models. Foundations of Artificial Intelligence Frameworks: Notion and Limits of AGI, which argues that current neural network paradigms are insufficient for genuine understanding and proposes a new framework for machine intelligence. Bridging Philosophy and Machine Learning: A Structuralist Framework for Classifying Neural Network Representations, which develops a framework for classifying neural network representations and highlights the need for a more nuanced understanding of the relationship between philosophy and machine learning.