As we move towards an increasingly enriched digital world, conversational Artificial Intelligence or Conversational AI I is being used to enable communication with that world. A dominant current example of this technology is chatbots or virtual assistants.
Over the last few years, there has been tremendous growth in the deep learning based conversation modelling techniques based on sequence-to-sequence modelling, transformer architectures and pre-trained large language models. Despite of these great advancements, these models are still far from being useful for their deployments in industrial settings and most of the current deployed chatbots use bot building frameworks such as Watson Assistant, DialogFlow and lex. These frameworks provide easy ways for modelling user intents using machine learning methods and using a rule-based dialog flow. Building chatbots using these frameworks could take a long time as they may require a lot of human involvement.
The deep learning based frameworks are completely data-driven and do not require involvement from domain experts. On the other hand, bot-building frameworks require a lot of human involvement. Despite the huge cost and time involved, enterprises use bot-building frameworks as they provide good control for what the end users experience.
At IBM research India, we are working on various approaches that could help bring the deep learning based models closer to their use in enterprise settings.
Grounded Dialog Response Generation
One of the key problems that deep learning based systems based on pre-trained large language models suffer is the problem of hallucination. They could provide factually incorrect or inconsistent information. To overcome this problem, we are exploring ways in which we ground the dialog response generation on trustworthy enterprise content. We are exploring methods in which we could use approved unstructured documents while response generation. We are also exploring ways in which we could use decision trees or flow charts as well as structured data for grounding dialog responses.
Human in the Loop for Improving Modeling
In order to bring more trust while deploying deep learning based models in conversational systems, we are exploring ways in which systems built using these technologies could be deployed in a controlled environment and could be improved as they are used in that environment. One natural place for achieving this in the case of customer care is to use human agents in contact centers. We are exploring deep learning based methods that could be used to generate responses given customer queries as well as recommend documents or small textual chunks that could be used to solve the given customer problems. We are also exploring incremental learning methods that could use the interactions done by human agents on these recommendations to further improve the systems.
Bootstrapping and Continuous Improvement of Chatbots
For building chat-bots using frameworks such as Watson Assistant and DialogFlow, a dialog designer spends enormous time in figuring out what problems are going to be asked by the bot and how should they be responded and then modelling them in the form of intents, entities and dialog flow. We have been exploring how we could use past human-to-human conversations and learn these model artifacts automatically. We have also been exploring explainable deep learning techniques for conversation modelling.