.png){kind=icon}
.png){kind=icon}
Return
Machine understanding
.png){kind=icon}
OA
AI
DOI：10.1016/j.tics.2026.04.003.png)
.png){kind=icon}
.png){kind=icon}
Share.png){kind=icon}
Save
Abstract
En 中文
New developments in artificial intelligence seem to be expanding the scope of what machines ‘understand’: image recognition systems appear to possess some understanding of objects and scenes, while large language models appear to possess some understanding of language. Evidence of advances (or limitations) in machine understanding is used to make claims about the safety and intelligence of artificial intelligence systems. However, such claims require an account of ‘machine understanding’ that clearly specifies what constitutes understanding and how it can be evaluated. While the fields of artificial intelligence and machine learning have not converged on an account of machine understanding, different assumptions are reflected in contemporary practice and relate to accounts of understanding from philosophy and the cognitive sciences.
Keywords:
Understanding
world models
benchmarking
AI evaluation
interpretability
AI Summary
Key information extracted from the uploaded paper, including a brief overview, abstract, background, key highlights, visual analysis, and future outlook.
Journal
IF:
17.2
Papers:
3.6K
Citations:
3.5W