IBM is analyzing public opinion on topics for the new “That’s Debatable” show on Bloomberg Television after upgrading the natural language processing (NLP) of its Watson AI. The new Key Point Analysis technique identifies opinions and key arguments from any number of sources before combining them all together into a coherent single story, with implications for businesses far beyond a television show.
Debatable AI
“That’s Debatable” is a limited series from Bloomberg and IBM where experts and leaders in business and policy take sides on a major question. For the first episode on the motion that “It’s Time to Redistribute the Wealth,” moderator John Donvan (virtually) hosted former U.S. Labor Secretary Robert Reich and former Greece Finance Minister Yanis Varoufakis to argue in favor, with former U.S. Treasury Secretary Lawrence Summers and Manhattan Institute Senior Fellow Allison Schrager arguing against. Ahead of the airing, the public was asked to share their opinions and reasons on the topic, and more than 3,500 sent in entries.
Watson applied Key Point Analysis to them, sifting 1,600 arguments from the total, and 20 key points among them. Those points were then shared with the moderator and panelists, who used them during the show. The submitted arguments split 56% in favor of redistribution, and 44% against. Meanwhile, for the show, 57% of the virtual audience was on the pro side, 20$ were against, and 23% were undecided. At the end of the show, the voting shifted to 59% for and 37% against. The 17% jump in the group against the motion meant that side was declared the winner.
Key Points
Key Points Analysis is about more than television debates, but the show is a good illustration of the technique and how it can be used in many contexts. As language is both universal and highly personal, applying artificial intelligence to better clarify what people are saying and writing can enormously helpful for sharing information. Most recently, IBM started offering a free version of Watson to U.S. states to handle questions about the November 3 election. Watson is responding with text and voice to queries about where and how to vote and other logistics. Idaho and Louisiana have both taken up IBM on the offer.
COVID-19 has accelerated the adoption of Watson by many governmental agencies globally. In Seoul, South Korea, IBM worked with the city government to create a virtual assistant that can survey and analyze the impact of the virus on people’s lives. The information Watson collects will then be used to inform policy decisions by city leaders. Meanwhile, the National Health Service of Wales launched its own Watson-powered virtual assistant named CERi to answer the public’s questions, in both English and Welsh, while the Royal Marsden in England set up a Watson-based virtual assistant for use by employees. IBM points specifically to the business world as a place for the technique but is shopping Watson around much more broadly.
“Using Key Point Analysis, businesses can gain a clearer view of relevant points and considerations to help make data-driven decisions on important operational questions,” IBM explained in announcing the feature. “The use of Key Point Analysis in “That’s Debatable” builds on a series of actions that demonstrate how IBM is advancing Watson’s ability to understand the language of business to help companies generate new insights – from commercializing cutting-edge capabilities from Project Debater, to transforming the fan experience at the US Open and helping states get critical voting information to citizens.”
Follow @voicebotai Follow @erichschwartz
IBM Offers States Free Watson Virtual Assistant to Answer Voter Election Questions
Seoul Launches IBM Watson-Based Virtual Assistant to Survey COVID-19 Impact
British Hospital Launches Virtual Assistant Using IBM Watson AI