IBM has tested one of its latest pieces of research “Project Debater” which is an AI system that is designed to debate humans on complex topics. The AI system was challenged in a debate against the grand finalist in 2016’s World Debating Championships, Harish Natarajan. While Natarajan was able to beat the machine, it was only by a narrow margin. The topic of the debate was subsidizing preschool. Project Debater argued in favor of the subsidy while Natarajan argued against it. The audience was polled before the debate to determine their stance on the topic which was 79% in favor of the subsidy. After the debate, the audience was polled a second time, and 17% of the audience opposed the subsidy, switching their stance to support Natarajan’s argument.
While Natarajan won the debate, IBM’s Project Debater demonstrated that it was yet again able to accurately interpret and respond to more advanced dialogue. Afterward, the audience even voted that they learned more from the device than from Natarajan, which was one of IBM’s aims in designing the software. IBM developed this software to not only construct persuasive arguments but to help people make more informed decisions. Project Debater is able to form arguments because it leverages data from news outlets and can pull specific details that best support its claims.
Project Debater is currently the epitome of AI technology. Project Debater has broadened the extent to which AI systems can interact and understand us which begs the question, do you think that there will ever be an AI system that can defeat a human in a debate? If so do you think that we will see this advancement in our lifetime?