Humans are bestowed with certain levels of cognitive skills and coherency that allow us to present our ideas, opinions, and arguments. But is the supremacy of human intelligence about to be challenged? IBM says it has built an AI that can almost mimic these skills. Project Debater, a successor to Watson, is being touted by Big Blue as an autonomous debating system that can engage in a competitive debate with humans.
For several decades, researchers have been trying to build machines powered by artificial intelligence to support humans. Getting such an AI to argue has huge technical implications. It can be considered as pushing these machines out of their comfort zone, closely studying and adapting to human mannerism, and, to a certain extent, our intelligence. The ability to debate requires knowing facts, connecting the dots, and presenting clear, coherent, and understandable arguments.
Project Debater has already been tested on the battlefield against internationally renowned debaters and has also won a few competitions. Even though it is intelligent, Project Debater is a machine, making its win a remarkable achievement. It took IBM’s team of researchers about seven years to develop this AI. It has been built using an NLP or a natural language processing model and trained using deep learning and machine learning techniques.
In 2019, the AI debated one of our finest, Harish NatarajanOpens a new window , a global debating champion. They presented their arguments on the question of subsidies in preschool. Although the machine lost the battle in an audience vote, it impressed the audience with its logical opening statement, sound arguments, and presence of ‘mind.’
Learn More: 6 Trends Shaping Machine Learning in the Post-Pandemic Era
During the debate, both participants were given 15 minutes to collect their thoughts and did not access the Internet. While Natarajan wrote down his points of arguments, the Debater went through its library of pre-stored information, newspapers, Wikipedia, and other published articles. Both contestants had to deliver a 4-minute speech on their stand along with a two-minute closing statement.
In all the debates IBM’s AI Debater has participated in, it has made its mark as a formidable opponent. In a different debate, the Debater along with Slonim and fellow IBM researcher Ranit Aharonov was able to sway the opinions of nine people in its favor. However, the entire interaction was awkward in the least. The linguistic ability that a human can showcase is still beyond the grasps of a machine.
Meanwhile, researchers are working hard to predict the sentiment of a single sentence to more complex tasks such as machine translation and dialogue systems. Noam Slonim, an engineer at IBM Research and principal investigator of Project Debater, said, “In recent years there’s been a tremendous amount of work in developing algorithms that can understand and generate human language.â€
Although a debate between a human and an AI is fascinating, this is not the first case of human-machine contests. In the past, AI has competed against humans and won several video games and tournaments. On January 14, 2011, Jeopardy participants and audiences saw IBM’s Watson defeat two of their all-time best contestants. IBM’s Deep BlueOpens a new window also defeated chess champion Gary Kasparov in a six-match series in 1997.
Until recently, the focus of AI development was narrowly-defined tasks as these require fewer resources to pursue, have clearer evaluation metrics, and offer end-to-end solutions such as those stemming from the rapid progress in the study of deep learning techniques. With Project Debater, IBM has broken through this hurdle and made it possible for AI tools to tackle complex tasks associated with human cognitive activities and require the simultaneous application of multiple skills.
Learn More: 6 Ways to Build AI That Incorporates Integrity, Diversity, and Ethics
But researchers do agree that inherent human skills are challenging to mimic. AI can win some Go and Chess games, but when it comes to subjectivity, reasoning, interpretation, and linguistic complexity, humans prevail, at least for now. Big Blue has now released the architecture of its debater that can engage humans in competitive debates.
It cannot also be said for sure that, despite its impressive capabilities, Project Debater will find many takers in the real world. The technology can certainly sift through large data volumes to form arguments and deliver value, but can it be trusted to host business meetings and form business strategies based on historical insights? Will businesses also depend on it to develop employee-centric policies or form expansion strategies in future?
To gauge the real-world capabilities of AI, one cannot help but look at the recent history of Watson Health- IBM’s $1 billion AI-powered initiative to sift through large volumes of clinical data, healthcare research knowledge, and patient information to advise treatments and procedures.
Despite six years of intense efforts to make it a success, The Wall Street Journal reported in February that IBM was mulling the sale of Watson Health as “physicians were hesitant to adopt artificial intelligence.†Watson Health’s eventual fate became apparent in 2018 when IBM embarked on a major layoff campaign targeting the Watson division with a special focus on Watson Health. AN IBM spokesperson admitted then that Big Blue was looking to “move to more technology-intensive offerings, simplified processes, and automation to drive speed†and to “hire aggressively in critical new areas that deliver value for our clients and IBM.â€Â
Do you think Project Debater is equipped to handle real-world challenges? Comment below or let us know on LinkedInOpens a new window , TwitterOpens a new window , or FacebookOpens a new window . We’d love to hear from you!