IBM Watson’s New AI Tool Aims to Help Businesses Identify Client Needs

IBM Watson introduced a new advancement in natural language processing (NLP) during the debut episode of “That’s Debatable” on Bloomberg Television. IBM’s new Key Point Analysis technology was used to distill over 3,500 viewer submissions into an overview of the global public opinion on the statement, “It’s time to redistribute the wealth.”

Yisela Alvarez Trentini
4 min readOct 30, 2020

During the debut of “That’s Debatable” on October 9, IBM Watson presented their latest advancement in NLP from IBM Research.

The show is a limited series hosted by John Donvan, presented by Bloomberg Media and Intelligence Squared US, and sponsored by IBM. It features economists, intellectuals, industry leaders, and policy-makers who debate today’s most pressing issues.

The debut episode convened a vibrant debate that included former US Labor Secretary Robert Reich and former Greek Finance Minister Yanis Varoufakis, both arguing for the notion that “It’s Time to Redistribute the Wealth.” Manhattan Institute Senior Fellow Allison Schrager and former US Treasury Secretary Lawrence Summers argued against it.

In order to determine who had won the debate, the virtual audience of the show was polled before and after the show. Prior to the start, 57% of people polled were for the motion, 20% against, and 23% undecided. After the debate, the number had changed to 59% for and 37% against. With an increase of 17 percentage points, Shrager and Summers were declared the winners.

The use of Key Point Analysis in “That’s Debatable” was able to generate new insights based on the analysis of submissions by the public, identifying over 20 key points in 1,600 selected arguments.

Key Point Analysis

The Technology

Key Point Analysis is a novel advancement in NLP developed by IBM Research. This next-generation NLP-based extractive summarization was evolved from the earlier IBM Project Debater, an artificial intelligence (AI) system that can debate humans on complex topics.

The new technology was used during the show to determine the main points that mattered most to the public based on the over 3,500 submissions made online before the debate. To do so, it utilized four steps.

First, the system classified arguments using a deep neural network to determine if the content of the submissions was for or against the motion. Irrelevant and neutral comments were removed. The quality was then evaluated for each argument, identifying potential key points through grading and filtering those with the highest quality. Points that were too long, emotional in tone, or incoherent were disregarded. Finally, arguments were identified for each potential key point and submitted with a prevalence for each. A small subset of the strongest arguments was used to create silent narratives arguing the pros and cons of the debate.


From the total of submissions, 1,600 arguments and 20 key points were identified.

About 56% of arguments analyzed favored the motion to redistribute wealth, while about 20% affirmed there was too much global wealth inequality. One of the arguments identified was that income inequality had dramatically increased over the past few decades and had caused suffering to many people — a tendency that would continue if wealth was not redistributed.

The remaining 44% of arguments were against redistributing wealth. From them, 15% argued this could discourage people from working hard, with examples cited about negative effects in entrepreneurship, individual initiative, and accountability for choices.

About IBM Research

IBM Research has been propelling innovation in IBM for 75 years. With more than 3,000 researchers across the globe distributed over a dozen locations, its mission is to positively impact business and society through science.

The Future of Key Point Analysis

Key Point Analysis is designed to empower businesses to employ AI for greater accuracy and efficiency. This can result in less data consumption and human oversight while giving companies a clearer view of relevant points and considerations so they can make data-driven operational decisions such as adjusting prices, evolving products, creating new marketing campaigns, and optimizing inventory.

The use of Key Point Analysis during “That’s Debatable” demonstrates that IBM is advancing Watson’s ability to understand the language of business — one that usually presents its own vernacular and is constantly evolving in response to new innovations, world events, and consumer expectations.

IBM has plans to commercialize Key Point Analysis as part of Watson NLP products. This could include commercializing cutting-edge capabilities from Project Debater, helping states deliver critical voting information to citizens, and even transforming the fan experience at the US Open. For now, you can see it in action during the next episode of the show, which will cover whether “A US-China Space Race Is Good for Humanity.”

This article was originally published in Startup Savant on October 30, 2020. Link:



Yisela Alvarez Trentini

Anthropologist & User Experience Designer. I write about science and technology. Robot whisperer. VR enthusiast. Gamer. @yisela_at