March 11, 2020 | Written by: Rob Thomas
Categorized: AI | IBM Research
Share this post:
IBM Research’s Project Debater competes in a public debate with champion debater, Harish Natarajan, on Feb. 12, 2019, in San Francisco.
There have been several seminal moments in the recent history of AI.
In the mid-1990s, IBM created the Deep Blue system that played and beat world chess champion, Garry Kasparov in a live tournament. In 2011, we unveiled Watson, a natural language question and answering system, and put it on the hit television quiz show, Jeopardy!, to compete against human champions, Ken Jennings and Brad Rutter. Last year, we pushed the AI boundaries again with IBM Research’s groundbreaking innovation, Project Debater. In a series of public debates with champion debaters, we demonstrated how the system’s advanced natural language processing (NLP) capabilities enabled it to debate humans on complex topics, in real-time.
In all three “man vs. machine” contests, we pushed the technological frontiers of what’s possible and inspired people to imagine new possibilities for AI.
Today we’re excited to once again bring technologies that were born in IBM Research to market – the first commercialization of Project Debater technologies, integrated into Watson. With new capabilities we intend to empower organizations to do things with AI they haven’t done before.
For example, a new sentiment analysis capability, being integrated into Watson Natural Language Understanding, is designed to help businesses begin identifying, understanding and analyzing some of the most challenging aspects of the English language – like idioms – with greater clarity, for more informed insights. Phrases, like ‘over the moon’ or ‘hot under the collar,’ have been challenging for AI systems because they are difficult for algorithms to spot, let alone understand and analyze. With new capability, businesses can begin getting deeper, more accurate insights from their volumes of language-base data, such as text data from call center operations and chat bots.
A capability like this aligns with our approach to AI, which simply put, says that AI is only as good as the data that’s fed into it. Consider the call center example. If your AI is unable to sense the emotion or the sentiment in the language your callers are using when they correspond in a chat, you must question the veracity of the outcomes. Call centers are one of the few customer points of contact. Understanding if a caller is positive or negative and why, can be critical to being able to improve the operation.
That’s just one of several key NLP enhancements we’re bring to Watson. The specifics of the news can be found here.
As AI adoption continues to rise in businesses around the world and across industries, consideration for the aspects of data capture and analysis will grow more critical. For IBM, the field of AI is about data and AI, and they are inseparable. As more data, structured and unstructured, is captured and funneled into AI, NLP is expected to play an increasingly critical role throughout the year, helping businesses not only better understand existing data, but consider new pipelines of data for analysis.
Register here to view a webinar on the news which will be broadcast on Wednesday, March 18, at 1:00PM ET.