Tab navigation
Competing at Jeopardy! is just the first step
As difficult as it will be for the Watson computing system to compete on the leading quiz show for brainy humans, the ultimate test will be to move beyond the Jeopardy! Challenge. The goal is to have computers start to interact in natural human terms across a range of applications and processes, understanding the questions that humans ask and providing answers that humans can understand and justify.
The DeepQA project at IBM shapes a grand challenge in Computer Science that aims to illustrate how the wide and growing accessibility of natural language content and the integration and advancement of Natural Language Processing, Information Retrieval, Machine Learning, Knowledge Representation and Reasoning — along with massively parallel computation — can drive open-domain automatic Question Answering technology to a point where it clearly and consistently rivals the best human performance.
Moving beyond the Jeopardy! competition, automatic Question Answering will help drive the future of business intelligence, analytics and information management, so that business and government decision makers will have the most cutting-edge capabilities for finding the precise information they need from the mountains of data they produce.
Why Jeopardy! is a great showcase for DeepQA
Onstage, completely isolated from other computers, the Internet, or human help, the Watson computer will receive the clue electronically, precisely when the human players see it. Competing in this way on Jeopardy! makes a great avenue for comparison between humans and machines. Jeopardy! is an excellent testing environment, in fact.
- Answering a Jeopardy! clue involves quickly focusing in on the most relevant parts of a clue, instantly identifying and dismissing non-essential or irrelevant parts, and then finding and weighing hundreds or thousands of bits and pieces of evidence to deliver a precise and correct answer.
- The best Jeopardy! players, according to our analysis, are able to do this over vast domains of knowledge and deliver the correct answer better than 85% of the time.
- If a player buzzes in and gets the question wrong, he or she is penalized. Winning, therefore, demands the ability to "know what you know" — to, in effect, determine an accurate confidence in whether or not you know the correct answer.
- While a natural and innate ability for humans, a challenge for a winning computer system is to consistently and very rapidly decide whether or not it has a correct understanding of the question and high enough confidence in its answer, based on the totality of what it has read, to take a chance on buzzing.
Related links
- Open Advancement of Question Answering - Learn about the Open Advancement of Question Answering initiative, spearheaded by IBM and CMU.
- Human versus Machine - More about the Human versus Machine contest.
- What about Web Search? - Is web search enough to win Jeopardy!?
- Deep Blue - A discussion of how this project compares to Deep Blue, IBM's system that beat a chess grand champion.
- FAQs - Frequently Asked Questions for Watson, the DeepQA project, and the Jeopardy! Challenge.
The DeepQA Project
The DeepQA project at IBMResearch is helping to make computers smarter in their interaction with people
