Difference between Watson and AlphaGo?...’Decision making support vs Self decision making’
Lee Gyung Tak | kt87@ | 2017-03-17 10:10:57

[Photo 1] : Watson (IBM) and AlphaGo (Google) are the most well-known and outstanding systems in the world of artificial intelligence (AI). Photo by IBM

IBM’s Watson and Google’s AlphaGo are in a fierce competition as they are the most well-known system in the world of artificial intelligence (AI). IBM is one of the companies that invested in AI development for a long time. Watson is able to effectively analyze unstructured big data such as words, texts, and pictures to find meaningful information and provide it for human use. Especially, Watson has advantages in voice recognition and natural language handling. Google’s AlphaGo AI engine can be used in various fields beyond Go, and Google``s AI technology has already been applied to Google services such as Google translator, photo, map, cloud, and smart car. In recent years, attempts have been made to commercialize AI in a variety of industries, including healthcare.

Industry experts have been asked whether IBM``s Watson and Google``s Alpha are better and their answer was that comparing Watson and AlphaGo is like comparing Apple and Orange. Which means it is difficult to make direct comparisons as they all have different purposes and different application areas. However, it is a general assessment that Watson is ahead in the commercialization part and AlphaGo is ahead in the academic field.

[Photo 2] : Google's AI engine can be applied in a variety of areas beyond Go (Baduk), and Google's AI technology has already been applied to Google services such as Google Translator, Photo, Maps, Cloud, and Google Car. Attempts have been made to commercialize AI in a variety of industries, including health care. Photo by Google

The big difference between the two AI applications is that IBM Watson focuses on ``decision making support`` to help people make better decisions, while AlphaGo has ``self decision making`` function. If we look at the Go (Baduk) game against Lee Sedol, Alpha thinks and judges by itself and makes its next move. This means, AlphaGo makes its decision by itself. However, IBM Watson analyzes physiological signals and images of MRI, CT, etc. in hospitals and suggests to physicians how to treat suspected diseases and treat them effectively. The decision itself is taken by doctors.

Another difference is the difference in computer hardware configuration that the AI operates. IBM is primarily based on one supercomputer, while Google is based on distributed computing with multiple smaller computers such as PCs. Google is reportedly trying to introduce a new quantum computer for AI. However quantum computers have more data to process at one time than ordinary computers and the processing time, on the other hand, is shorter.

The both companies are training AI through deep learning of machine learning technology. However, IBM’s Watson is applying human-made materials and papers to educate, and Google AI is learning based on massive Internet resources.

IBM’s Watson is getting a personalized education using specialized materials like medicine and law in certain fields such as medicine and law. Watson can create its own database with human-processed data and is able to use it to search for answers. Watson, on the other hand, has the disadvantage of not being able to answer questions that is not included in its data base.

IBM is trying to secure the data necessary for AI training. The company acquired the ‘True bell Health Analysts’, a health information database company in 2016 for KRW 3.2 trillion in order to secure 600 petabytes of medical data. In contrast, Google is able to acquire a large amount of user data from search engines, Android smartphones, and YouTube. Therefore, Google is training AI through pattern recognition with rich internet data. Cat photo recognition training can be one of the examples of Google AI training. By providing many cats photographs as data, Google AI recognizes the cat characteristics by itself so that it can read the cat pictures more accurately than people later.

The difference between the two companies approach to AI may be that the target customers of the both companies are different. In other words, while IBM focuses on Business-to-Business (B2B) and trains Watson to specific needs in response to customer needs, Google, on the other hand, is primarily focused on Business-to-Consumer (B2C). That is why Google uses more data than custom technologies to develop more generic technologies.

It is not important that whether Google is better or IBM is better as it is not easy to compare these two AI services. Obviously, these two US companies are leading the development of AI technology, and the AI, they are developing is increasingly being used in our daily lives. Although there are concerns that AI is replacing people and eliminating jobs it is undeniable AI is making our daily lives easier.

By Lee Gyung Tak kt87@

Source=Korea Evaluation Institute of Industrial Technology

[ copyright ⓒ The Digitaltimes ]