IBM Watson – Meet the World’s Most Famous AI
Remember a few years ago IBM developed a very clever Artificial Intelligence (now known to IBM as Augmented Intelligence) software that took part in the popular American game show Jeopardy? Remember it actually won the thing among some of the best Jeopardy players too? Of course you do (you don’t? Does not matter). It was the beginning of the reign of artificial intelligence on this world as we know today (no, not in the Terminator kind of way). It was a sort of Revelation moment for the world of technology and an actual artificial intelligence we can interact with. It was also a sort of glimpse into a future, a kind of what it could be like. At the end of the day though, it was more like the coolest thing in the world ever.
The idea Artificial/Augmented Intelligence (we will be using Augmented Intelligence [AI]from this point on for the benefit of the guys in IBM) is not something now. In fact the idea has been manifested into our heads ever since we watched the first of the Star Wars films. The idea that a robot has its own conscience and critical problem solving skills are all concepts of AI and machine learning. The idea is that a machine, or robot entity is able to determine a problem and solve it all on its own within its own programming. That particular idea manifested plenty of concepts and films including the Terminator series where we technically see what happens when AI goes wrong and rouge (Will Smith’s I, Robot for that matter shows what its like too). In the Iron Man cinematic universe we have J.A.R.V.I.S. (Just A Rather Very Intelligent System; if you are wondering, yes we are very big nerds) is a form of AI that could closely represent what IBM have built.
So, putting all these other fictional AI aside, the IBM Watson then. We got the pleasure of spending time with IBM’s baby for about an hour and it was amazing. What’s best about the whole encounter is that we only needed to speak to it through Telegram, a secured chat service with open source API. At this point I should mention that they did consider using Whatsapp as a means for us to communicate with Watson (Whatsapp Group). But due to the lack of necessary API from the iconic green coloured app, they went with Telegram as it was a more flexible platform.
The IBM Watson can be considered, if you might a rather clever ChatBot like no other. Like Siri and Google Assistant, you can speak to it and it will give you responses based on you requests and its library of programmed responses. Unlike Siri and Google Assistant however the IBM Watson has been developed since 2011. It was first developed to beat Jeopardy and beat the game it did. That won it US$ 1 Million. So technically it was built for fun. But in 2013 IBM decided to code Watson for enterprises instead. The first industry IBM decided to enter with Watson was the healthcare industry. Why? Because Watson can give its users solutions on-demand. It helps nurses and doctors make the right decision at the right time. That also means less mishaps, and more lives saved. IBM Watson today is so much more than just a medical assistant though.
Now for the technical bits. Watson is a little unique compared to Google Assistant, Siri, and even Microsoft’s Cortana. Watson employs something IBM calls a DeepQA. DeepQA, for those of you who are interested is an IBM developed operating system that is based on SUSE Linux Enterprise Server 11. That software is placed on 10 racks of IBM’s very powerful Power 750 servers to ensure Watson can answer even the most complex of equations and questions within a heartbeat. Those 10 racks of servers pack about 2,880 cores and 15 Terabytes (TB) of storage for its 80 teraflops ability.
As keen-eyed readers might point out as well Watson is indeed open sourced with its Linux-based operating system. In fact, some of the world’s fastest and most reliable super computers, like Watson run on a form of Linux Enterprise. Using its Linux-based software algorithm it could accurately pin point answers that would eventually win it the Jeopardy title. In this sense; unlike Google, Watson is made to give you precise answers as and when you need them. You might want to liken it tot Google’s “I’m Feeling Lucky” button but from where I was sitting, Watson seemed to work so much quicker.
At the same time, the speed may not overwhelm you; it is not like there is that much of a difference in response time anyway. But you have to take this into account; the thousands of cores on IBM’s Watson super computer does a database and internet search of over a hundred possible answers to a question anyone asks of it. Immediately after that the super computer kicks up to 200 algorithms to the possible answers to analyse different features and keywords from them. Those algorithms will pick the best, or most important answers and later presents a single answer that fits the algorithms’ rankings and confidence levels. When you put that into perspective it sounds all the more amazing.
Which brings us to the day we actually met the IBM Watson. If we have to tell you, we would not have to kill you. We were all sat down in press conference style meeting hall and then was briefed on what Watson meant to the IBM team and what they have done to bring Watson to this day. In fact the demonstration on this day is possible due to the works of a bunch of coders grown in-house by IBM’s Bluemix Cloud Platform workshops. We were told that the session by IBM was to show what Watson could do. Of course we were allowed to ask it all sorts of questions.
It sort of works like a highly advanced chatbot that answers you properly rather than giving you sarcastic answers. We even have to do it through the Telegram app and give it specific commands. Obviously we could do it with voice activation either and Watson (in a woman’s voice I have to add) responds with whatever answers it found off the web, or if it could not understand your question it will let you know. So that was what we did, we asked a bunch of weird questions and Watson immediately responds with all sorts of information that we needed. Of course, the more specific your command the more accurate its response would be.
On the surface then, Watson is pretty much like any other machine learning AI like Apple’s Siri, Samsung’s S Voice, Google’s Assistant, Microsoft’s Cortana, and Amazon’s ECHO. But we were all amazed by one thing that IBM’s Watson does; its response time. With all the other AI, when you tell it to do something there is a waiting time, or processing time if you might. The server would have to try to understand your questions first and then put them into ones and zeroes for it to understand and break the question down. That gives all the other AI assistants a buffer time that extends from a few seconds to a minute, or two in some cases. With Watson though, every question and command was met with a near instant response. I would say that it was instant but at the same time there is always a small amount of near negligible response time. It was amazing; it was like looking into the future of machine learning in some sense.
The developers that is maintaining the Watson algorithm (some local guys of course) says that Watson is still learning and adapting. They do indicate that Watson is constantly improving while learning to adapt to its specific usage patterns in a single corporation or by a single person. They also said that Watson is not perfect just yet; and on that, we agree. It did miss some simple questions by the developers and finds some questions a little incomprehensible. After repeating the questions (by voice) though Watson did answer all of them correctly without much of a hiccup.
The demonstration was just a glimpse of what Watson can do though. The integration with Telegram was just an example of how Watson can actually be used on a daily basis just like how Google’s Assistant work. It is just that much faster and sleeker though in our perspective. It was also a glimpse to a machine learning future. Examples? General Motors is currently deploying Watson to create OnStar Go to provide personalised contents from their favourite brands through their vehicles. Thomson Reuters is using Watson to analyse trends and therefore allowing the company to accurately advise its customers. Even Autodesk is utilising Watson to improve its customer satisfaction rating by allowing its customers to communicate and improve their experiences in real-time.
The use of AI is not going away anytime soon. As a matter of fact, if you are a gamer and have played games like Halo; you can expect to see highly advanced AI like the ones you see in-game. Machine learning is new but it is growing at a rate faster than ever which also means AI grows even more intelligent as days go. Whether we like it or not AI is going to play a bigger part to shape our daily lives in the future and Watson made the future look good. it still does have its flaws. Those flaws though are being rectified on a daily basis not just by IBM’s team. Other manufacturers and solutions provider is also working hard to give us the AI we deserve obviously. But as of the session; not many can come close to it at this day.
Source: IBM Malaysia