According to Amazon, “Now, we’re taking another step toward natural interaction with a capability that lets Alexa infer customers’ latent goals — goals that are implicit in customer requests but not directly expressed. For instance, if a customer asks, ‘How long does it take to steep tea?’, the latent goal could be setting a timer for steeping a cup of tea. With the new capability, Alexa might answer that question, ‘Five minutes is a good place to start’, then follow up by asking, ‘Would you like me to set a timer for five minutes?’”
Amazon notes that while the concept might seem simple on the surface, they say that under the hood, “sophisticated algorithms” are being used to determine if a question might need a followup. This is done to help prevent the feature from being too disruptive to users, where it will try to determine if whether or not your query warrants a followup or if it doesn’t.
Amazon will also rely on what is known as bandit learning to help the AI track whether or not the recommendations are helping, and for those that aren’t quite as useful, they will be suppressed.