If you come to think about it, the human-computer interaction (HCI) is one of the most happening areas for any tech-crazy nerd. What started as a command-based, on screen interaction has today evolved into a more natural collaboration between man and the machine. Artificial intelligence (AI) surely is seen by many as the next big step in almost every discipline of commerce and industry. Though this does bring in the bloated expectations which have already been built around this technology, it has also opened up the much-needed discussion on broadening the scope of HCI.
HCI : Adding the human in the machine
There is no denying the fact that the AI-human relationship will reach its full potential once the computers interact as naturally as humans interact with each other. Effective conversation with context, thus, is not an option but an essential element of the AI of the future. The move towards conversation with context has already begun with our phones and homes having AI-powered virtual assistants. These chatbots enable the users to ask conversational questions and receive answers in text form. Though a major hit in the form of smart appliances and customer support applications, users want more interactive interaction with these AI-powered assistants.
The critical elements in natural learning processing (NLP) are speech recognition and transforming text into speech. As machine learning undergoes significant advances, it is expected that conversational systems will be able to recognize speech in a more precise manner. But, the catch is – these conversational bots can only respond human-like when specific queries are raised. To expect them to strike a one-to-one human conversation is a high expectation to have this early in the game. The conversational agents have not even evolved enough to sense and respond to emotions – the users on their part however, expect copious amounts of empathy from the machines.
Such requirements and exceeding expectation from the end-users have pushed up the need to increase the human element in machines. The major reason behind this limitation is the fact that our computers are yet to grow their ability to understand natural human language. Also, the possibility to anticipate the way in which a conversation might grow makes it even more difficult for engineers to script an automated response.
Priority One : Bringing context to the conversation
It is in this situation of growing expectations, lack of proper response, and the resultant confusion, that ‘conversation with context’ has gained so much importance. The context here is a very broad term and involves multiple layers of knowledge and information. Adding context to the conversational systems, as challenging as it is, can be achieved by synergizing machine learning algorithms along with supervised learning and reinforcement learning. While supervised learning involves teaching computers through examples, reinforcement learning helps in mapping a context and the associated response accordingly.
As all these elements are brought together on a common platform, it is expected that the ground and context of the conversation will broaden up immensely. It is also expected that apps will pick up signals, emotions, and gestures, adding more emotional intelligence to the entire interaction with humans. Once AI reaches the point wherein it can truly converse with humans in the most interactive manner possible, the possibilities of it revolutionizing the realms of healthcare, personal care, education and businesses, will be the natural next step. Until then, the learning curve shall continue.
