GPT-4 has a great potential to generate Cypher statements based on only the provided graph schema, leveraging its extensive knowledge of various datasets and graph models during its training. However, limitations include issues with multiple aggregations using different grouping keys, version two of the Graph Data Science library being beyond its knowledge cutoff date, occasional relationship direction mistakes, non-deterministic behavior, and sometimes providing explanations despite instructions to avoid them. Despite these limitations, GPT-4 can be a valuable tool for experimental setups or when examples of Cypher statements are provided, offering impressive generalization ability in some cases.