ChatGPT is a versatile tool that generates natural-looking texts, with a surprising narrative and conversational fluency. However, it has limitations: its data ends in 2021 and, as it does not have access to the internet, it lacks updated information, a barrier that is insuperable today in a quick comparison with any search engine.

Moreover, it does not always find reliable sources or robust evidence, and it is this —and not the apparent coherence of the text— that determines the veracity or reliability of an argument.

Artificial intelligence is valid for situations that hold a certain margin of error, even some nonsense. But it is not valid for critical issues, such as scientific work, legal or financial advice, or medical consultations. It produces a misleading illusion of rational thought, but it has no reasoning or reliable knowledge about the world. It does not understand, in a human sense, anything it writes.

Full article