there are several key limitations to using ChatGPT for customer support:

1. Limited understanding of context: ChatGPT can struggle to grasp complex contexts, nuances, and subtleties in customer queries, potentially leading to inaccurate or inappropriate responses.

2. Lack of empathy: As an AI model, ChatGPT lacks true emotional intelligence and cannot provide the human touch and empathy that is often crucial in customer service interactions.

3. Fabricated information: ChatGPT can sometimes generate "hallucinations" or make up fictitious information and present it as fact.

4. Inconsistent answers: If not trained on comprehensive datasets, ChatGPT may provide inconsistent responses to the same questions, which can damage customer experience.

5. Inability to handle complex queries: ChatGPT struggles with intricate, deeply technical, or unique scenario questions that require specialized knowledge and critical analysis.

6. Security and privacy concerns: Using ChatGPT involves processing and storing sensitive customer data, which raises potential security and privacy issues.

7. Outdated information: ChatGPT's knowledge cutoff date means it may not have the most up-to-date information, potentially leading to inaccurate responses.

8. Potential for bias: ChatGPT can sometimes offer unintentionally biased or discriminatory responses that may harm a brand's reputation.

9. Need for human oversight: Due to these limitations, ChatGPT requires constant supervision and management to ensure proper functioning and to correct errors.

10. Limited integration: Without proper integration with existing customer service systems, ChatGPT can lead to workflow disruptions and isolated information.

Given these limitations, it's generally recommended to use ChatGPT as an internal tool to support human agents rather than directly interacting with customers, at least for now.