The hottest ibmwatson expands smart platform to im

  • Detail

IBM Watson expanded the intelligent platform to improve human-computer interaction

with artificial intelligence (AI) machines, what should be the IQ? Can artificial intelligence replace human beings? At present, no one can give a positive answer, but machines make judgments through learning, and human beings complement each other with their experience, which is expected to achieve the best combination

with artificial intelligence (AI) machines, what is the IQ

ibm Watson's 2011 TV program jeopardy Defeating the two champions in the puzzle game triggered a heated discussion on AI technology and brought a lot of imagination. Whether the human brain is smart or the computer through training remains to be further developed by artificial intelligence. From the perspective of commercial market, IBM Watson shines brightly in medical, financial, retail and other markets with its unique cognitive computing ability. The technology behind it is not just data analysis. Through its personality insights technology, it analyzes written text to understand users' personality and preferences, and uses natural semantic analysis technology to achieve near ideal human-computer interaction

dev mookerjee, technical director of IBM Watson Asia Pacific, introduced IBM Watson and its latest development in artificial intelligence during his recent visit to Taiwan. At the beginning, he emphasized that artificial intelligence enables human beings to interact with machines, such as the currently common chat robot (Chatbot). The difference between machine and human is that machine is good at reading rows and columns and not good at intuitive thinking, but human beings are on the contrary, and AI just integrates the two worlds

dev mookerjee, ibm2, Watson Asia Pacific Technical Director of joint strength experiment

watson AI what's special

if you used AI in search engines ten years ago, you might not have much information, but now you will find dozens of manufacturers and different application fields. Facing the rapid increase of new competitors, mookerjee said: we started early in the development of AI technology and are developing technology in an all-round way. In addition to enhancing deep learning ability, it will also create new chips to simulate the operation mode of human brain

The Watson system tries to operate in a human cognitive behavior pattern. Mookerjee pointed out that we train Watson cognitive system to simulate the mode of human thinking of automotive materials entering the aluminum age, and train how to learn and make decisions like humans. Watson is a probabilistic system, which improves its confidence in answering questions through training

the trained Watson system can understand natural language, understand and analyze the content of the customer's speech through cognition, rather than just Judging according to the set rules. Mookerjee cited the use case of an Australian insurance company on automobile claims. After the insurer fills out the claim form, Watson will read these data first, and then judge whether there is fraud through language use, so as to assist the insurance company in the audit work at the first time

chat robots that can understand natural language can not only answer questions, but also understand the interlocutor's intention through language understanding ability. Mookerjee takes booking tickets as an example. General chat robots need to ask and answer questions step by step with interlocutors to complete the booking process. Watson's Chatbot technology tries to understand what interlocutors want, so that users can book tickets in their usual way, not just a question and answer dialogue. If you want to further recommend the best price, you can integrate more with the rear system to provide decision-making solutions. Mookerjee said: data analysis is Watson's strength, but it is only one of its functions, but it also has serious wear on the rotor. We build this system to understand the way humans talk and ask questions, not just answer questions

mookerjee mentioned that personalityinsights is a unique technology of IBM, which can analyze the personality and preferences of consumers by writing words such as letters, blogs or road articles, understand whether this person is relatively open to new things, and then analyze their possible decisions in purchasing behavior

machine learning and augmented intelligence

mookerjee believes that IOT is one of the good application fields. IOT collects a large number of machine generated data through machines such as sensors, most of which are structured data that can be used for analysis. However, there are still many unstructured data in IOT, such as sound and images taken by cameras. Moo called it "newclear" kerjee stressed that collecting and analyzing unstructured data is Watson's strength, so he can help deal with these raw materials. In addition, a special IOT application is to set up cameras at the final stage of production to detect whether the finished product packaging is damaged through images, without relying on manual inspection

in the past, these unstructured data such as images, PDFs and emails could not be accessed by some traditional systems, but Watson system can help enterprises read and use these unstructured data

speaking of the main application markets, the IBM Watson platform was first used in medical treatment, and then extended to financial, retail, government, telecommunications and other markets. Mookerjee mentioned that this system was not designed specifically for some applications at the beginning. Because these fields have a large number of unstructured data, and the core ability of Watson platform is to analyze and interpret these unstructured data, it is naturally used in these fields

doctors and financial professionals used to spend a lot of time studying huge medical reports or financial information, and Watson system can quickly find corresponding solutions through machine learning, analysis and interpretation, such as providing retirement plan consultation or quickly diagnosing diseases and recommending appropriate treatments. One of the most famous examples is that the University of Tokyo in Japan used Watson to diagnose a patient with rare leukemia within ten minutes and find out the appropriate treatment, because Watson system itself has read 20million cancer research reports, and the reading volume is quite amazing

customer service in the financial industry is also one of the fast-growing applications at present. Some banks have developed customer service robots using Watson system to naturally talk with consumers and provide a customer service system all year round. For example, Yushan bank in Taiwan adopts Watson cognitive technology, supports traditional Chinese natural semantic analysis technology through IBM Watson conversation, collects collective conversation content and carries out voice analysis and recognition, as well as continuous training and augmented intelligence, so as to provide intelligent product recommendation and financial consultation. AI technology's ability to understand and master natural language is expected to continue to improve, and relevant services may be seen in many industries that need front-line customer service in the future

asked which is a high growth field, mookerjee believes that the market changes too fast, and it is not easy to predict what is needed at this time next year. Mookerjee explained: in 2015, the content we talked with the bank was different from now

Mookerjee believes that the Asia Pacific region is growing continuously, but each country has different acceptance of disruptive innovation technologies, such as New Zealand, Australia, Singapore, South Korea and India. In addition, mookerjee also mentioned that promoting products and services through partnerships is a good model. He emphasized with the example of SK Telecom in South Korea that our services run on SK Telecom's servers, and sktelecom has a team to help us promote IBM's services

Watson's customer base is mainly large enterprises, and it also provides corresponding solutions to start-ups. Mookerjee said: we create chat robots that can understand natural language and understand different user needs through machine learning. For example, the cost of Chatbot is low, which meets the requirements of some start-ups; Large enterprises need their own proprietary APIs and databases, which are large-scale and can adopt more advanced versions. In contrast, other competitive schemes first target the group of start-ups, and then upgrade to large enterprises, which is different from our strategy

the future of artificial intelligence

mookerjee stressed that artificial intelligence will continue to exist. Like the digital transformation in those years, it will become the market standard (NORM), and the rapid expansion and popularization of artificial intelligence applications are far faster than the original digitization speed. For example, over the past four years, Watson's core capabilities have increased significantly, and Watson can be seen in all aspects of IBM's services

Mookerjee said: after I came to IBM to work, I found that there are many ignored and unused data in the world, which can be processed and used in large quantities through Watson AI technology. Today's market is changing, and he predicts that people will have more expectations for AI in the next five years

in the future, it will be the general trend for enterprises to adopt AI technology. In addition to letting enterprises know more about customers, it can also improve good interactive relations. In terms of business theory, reducing customer effort can help improve customer loyalty. Mookerjee cited the customer effort score to explain that many companies have never thought about how much effort customers need to spend to interact with enterprises, and artificial intelligence can provide enterprises with the best assistance in this regard

artificial intelligence expands the knowledge and ability of people or enterprises, which is why IBM calls artificial intelligence expanded intelligence. Can artificial intelligence replace human beings? At present, no one can provide a positive answer. When machines make judgments through learning, if the training data are insufficient or encounter conditions that have not been touched, there will still be errors. Human beings believe that it is the best combination through their complementary experience

Copyright © 2011 JIN SHI