Why is Voice conversation AI a hard problem to solve?

Currently, most voice AI assistants and functions are meant for simple Q&A-type interactions, basically delivering FAQs in audio format. Most current phone IVRs fail at self-service predominantly due to two reasons:

1) They are not hardwired for the right use cases

2) Natural language processing or NLP does not recognize enough words to act on, when commanded.

Many conversational voice assistants lack the domain-independent NLP capabilities to execute conversations on topics outside of the knowledge base they are trained upon, or react to commands that come non-linearly. Voice Conversation is AI a hard problem to solve because users won’t consider it to be the predominant self-service interface until it provides an experience that is superior to a website or mobile app. To get there, enterprises like Agara focus on building domain-specific skills addressing the most frequent caller requests, such as replacing a lost credit card.