This House Believes a Computer Can’t Have an Intelligent Debate. It turns out that it can.

The US giant, IBM, has developed Artificial Intelligence that enables the computer to argue, in depth, with a human being.

Project Debater has the ability to analyze more than 400 million newspaper articles in the time it takes to drink a cup of coffee. It doesn’t sound too far-fetched to be true when you consider that virtual assistants like Google Home and Alexa already answer simple questions without any trouble at all, although let’s face it, a conversation with Siri can be quite hard work!

In 2019, Project Debater debated against award-winning Harish Natarajan, at a conference in San Francisco. Although it lost, the audience was impressed with how well it argued and how much information it was able to retrieve and share. IBM started making it in 2011 and it’s the first AI system that can manage such complex interaction with people by digesting masses of material and then constructing a structured argument – not only can it present its points clearly and intelligently, but it can also disagree with you and give valid reasons for it.

Project Debater’s knowledge bank consists of 10 billion sentences; it doesn’t get trained on the topic it is about to debate but instead it reacts like a human by ‘thinking’ quickly. It has been programmed to start with an opening speech based on the debate’s topic and it can do this because it is very capable of understanding human language and any nuances (sometimes even people fail to pick up on subtleties in speech). Its speech is coherent, structured and evidence-based. After this, it will ‘listen’ to the opponent’s response, then process it and build a suitable rebuttal.

Artificial Intelligence scientists have been working hard to find a way to make computers able to ‘comprehend’ and ‘think’, in order to simulate human intelligence as closely as possible. This development at IBM’s lab in Israel is a stride in the that direction. In the 90s, another IBM super computer became the first machine to beat a top chess champion when it competed against the Russian grandmaster, Garry Kasparov. The really tricky thing though, is to make AI capable of understanding human emotion.

During its San Francisco debate, Project Debater was not only able to thoroughly research and present facts, when it came down to emotion, it managed to show empathy too, at least to some degree. It said, "While I cannot experience poverty directly and have no complaints concerning my own standards of living, I still have the following to share: Regarding poverty, research clearly shows that a good preschool can help kids overcome the disadvantages often associated with poverty."

Does this mean AI is getting closer to mirroring the human mind? In 1950, an excellent mathematician and computer scientist called Alan Turing, devised what we now call The Turing Test. He proposed that if a machine could ‘fool’ a human into thinking it was also human (by conversing like one), then we would have achieved proper artificial intelligence. What would Project Debater score in the Alan Turing test?

Further reading:

Garry Kasparov: Why the world should embrace AI - BBC Future

Project Debater - IBM Research AI

Slonim.pdf (

Israeli debate champs beat computer, just barely - The Jerusalem Post (

IBM develops artificial intelligence system that can debate humans - The Jerusalem Post (

Alan Turing: Creator of modern computing - BBC Teach