Self-service applications for customers and employees have consistently featured poor user experiences that garner disappointing uptake. But now, large language models (LLMs) and generative AI (genAI) enable chatbots and interactive virtual assistants (IVAs) that are smarter, more useful, and more conversational than before. These AI capabilities also enable faster application delivery versus traditional methods. The potential is massive, but these new technologies carry risks of hallucinations, jailbreaks, and bias. This report will help brands understand the benefits and risks of modern AI capabilities in conversational AI to make informed decisions on its use.