IBM plans to integrate several key components, such as natural language processing (NLP), from the AI system into its question-answering computer system Watson from March this year. The move is aimed at “enabling businesses to capture, analyse, and understand more from human language and … transform how they use intellectual capital that’s codified in data.”

The most important technology behind Project Debater is NLP, which is a branch of advanced AI aimed at enabling communications between man and machine by using human language. NLP remains a challenge due to the inherent nuances and subtleties of human languages.

“For the first time, Watson will be able to identify, understand and analyse some of the most challenging aspects of human language, easily, for deeper business insights,” IBM said in its release, adding that a feature called advanced sentiment analysis would help machines to understand and analyse idioms and phrases with harder-to-grasp meanings.

Hatched in IBM’s Research Lab, the AI-powered Project Debater scored mixed results in a public debate against award-winning human debating champion Harish Natarajan last year in San Francisco.

Natarajan argued against subsidising preschool education in the debate and claimed the victory but most members of the audience said they thought the machine did a better job in enriching their knowledge on the topic.

Noam Slonim, a distinguished engineer at the IBM Haifa Research Lab and one of the designers of Project Debater, said a machine has to search through a collection of billions of sentences in the system and choose instantly useful pieces of information that are relevant to the topic to support its arguments in a consistent manner.

When asked whether a machine can actually understand the purpose of human debates, IBM said that level of AI, known as artificial general intelligence (AGI), is still "several decades away, not available until at least 2050.”