Virtual Assistants like Alexa, Siri, Cortana, Assistant, etc., are designed to help people perform day-to-day tasks quickly, but do they perform tasks intelligently? How effectively do they use Artificial Intelligence (AI) technologies? I would argue that most of the virtual assistants that are widely used and available do not demonstrate much intelligence, which is not to say that they are not useful or they cannot be adapted to execute more intelligent tasks. Certainly, there are many use cases where they are incredibly useful – responding to messages safely while driving, for instance. However, the future of virtual assistants will require effective use of AI technology to support smart interaction.
The present day virtual assistants facilitate one-way communication where the user will ask a question or give a command before the assistant executes certain tasks. The voice prompt enables the assistant to execute the underlined tasks. What about two-way communication? Where the assistant starts the communication with the user; what are the characteristics of these smart/intelligent assistants? Let me share a fable that illustrates what I mean by intelligent assistance.
A long time ago there lived a king. Every day his barber would come and give him a shave. While shaving, the barber would hear the conversations between the king and his minister. Over time, the barber began to think to himself, “I have all the same characteristics the minister has, why should I be a barber and not the minister?” Therefore, he tells the king, “My lord, I have all the qualifications to be your minister, please allow me an opportunity to serve you in that respect.” The king knew that the barber did not have the necessary qualifications, but decided to give him the opportunity regardless, in hopes of showing him the difference. One day a ship appeared on the horizon and not knowing the motives of those aboard, the king sent his new minister to find out more about them. To discover the identity of the newcomers he goes to the harbor, enquires about who the visitors are, and returns to tell the king. Hearing the identity of the visitors however, was not the only thing the king expected to hear. So, he asks the minister, “Well, why are they here?” Not knowing the answer, he returns again to the harbor. This scene repeats itself 100 times, before the king has all the information he required. When the minister returned the last time, the king tells him “Now, do you see why my old minister was the minister? She would have made one trip and gathered all the information I needed, while it took you 101 times.” Finally realizing his shortcomings, he begs the king for his old job, and the king happily obliges.
The virtual assistants of today are like the barber in the sense that they require constant interaction with the user to perform tasks. What a smart minister would do is make one trip and then return to the king with all the information, because he or she would realize that the king does not simply want to know the identity of the new arrivals, but also their purpose, length of stay, etc. He or she would use the context of the situation as well as their knowledge of the king and his requirements to provide the right information and executing the right actions, thus providing smart interaction.
Virtual Assistant’s need to be proactive and context-aware to truly be able to make intelligent choices or suggestions that would add greater value. In the current state, the assistants are great at doing easy tasks such as setting an alarm, making a phone call, or even checking for conflicting appointments. However, wouldn’t it be great if the virtual assistants knew which of your meetings are likely to run late and accordingly schedule and/or reschedule the future appointments.
The AI research over the last 30 years on causality, reasoning, sequencing, inferencing, and knowledge representation including temporal reasoning and natural language processing are key to developing a true smart and intelligent virtual assistant. Based on my understanding, Google, Apple, Amazon are all working on designing the next generation of virtual assistants and Siri, Alexa, Cortana are only mile markers on a long road.
