Addressing Uncertainty in LLMs to Enhance Reliability in Generative AI