How are AI models tested and does it make sense?!!

The article tackles the controversial issue of testing large language models (LLMs) as though they were human. The author argues that, despite their impressive text-generating capabilities, LLMs cannot think or feel like humans do. It seems that society often attributes human-like qualities to them, which leads to misunderstandings in evaluating their capabilities. The article emphasizes that testing LLMs at a human level can result in misinterpretations of their outputs and the ways in which they can be used in various applications. Instead, the author proposes a more realistic approach to assessing their performance that takes into account their actual capabilities and limitations. The conclusion suggests that past experiences with technology should teach us to approach LLMs with a greater understanding of their nature.