Text summarization can be either extractive or abstractive. In the extractive text summarization, a language model needs to extract important sentences in the document and concatenate them to generate summary. Whereas in abstractive text summarization, a model can use novel words in the summary which are not present in the source document, so it will look like a human-written summary. This current work is an overview of various authors’ ideas to generate abstractive summarization. Also, this work discussed some of the problems identified in abstractive summarization, such as coverage, factual inconsistency, redundancy, hallucinated information, missing logical structure of statements, and flow of writing problems.