It looks like you're using an Ad Blocker.
Please white-list or disable AboveTopSecret.com in your ad-blocking tool.
Some features of ATS will be disabled while you continue to use an ad-blocker.
There has never been a better time to collaborate with bots - they are your new digital friends. Artificial Intelligence (AI) bots are getting to maturity to assist you achieve extraordinary levels of productivity and simply change the way you work. This is the kind of AI that has become the focus of attention. So what is AI, why now and what has changed? Every decade has had high expectations of AI and often times, when AI did not deliver, this was called AI winter. There has always been a rise and fall of AI's reputation. However, new technologies that have emerged in the recent past are driving the convergence of computing power, algorithms in the area of neural networks, machine learning and easy access to implementations in open source tools to make a new type of AI called 'Applied AI' possible. Most importantly, all of this is driven by business needs to process unstructured information, learn, automate, predict and anticipate. This is what is called Applied AI - AI primarily focusing on the use of advanced machine learning and knowledge engineering techniques to build smart machines (software robots) that act like people. So what does it take to build such AI applications? How do we build machines (software robots, bots) that act like human beings? How do we imitate, automate, and enhance the abilities of the workforce for higher productivity is our pursuit of Applied AI. If we have to mimic - simulate through understanding the human action - we surely need to understand the human mental abilities and action. And these set of mental abilities are called cognition. Essentially AI requires abilities of human cognition. So are there any technologies that can implement cognition? If so, these technologies will be the basis for building AI applications. With that said, it is then imperative that cognition must be computable. Computable simply means that there should be algorithm(s) and data structures with some representational techniques that can compute any of the cognitive functions such as believing, consciousness, emotion, language, learning, memory, perception, planning, problem solving, reasoning, representation, categorization, concepts, mental imagery, sensation, thought, etc. So how difficult is it? Is Cognition actually computable? Not all but some aspects of cognition can be computed today using computational linguistics, computing statistics, computational geometry, and computational learning. There are also aspects which are not really computable like consciousness, believing, thought emotion etc. Wipro has a specific interest in enterprise IT. A study of the key use-cases led to the simple finding that the key cognitive capabilities required are primarily in the areas of Perception, Learning, Reasoning, Representation, and Language. If we could implement these functions in a software stack, we could then have an 'Applied AI Platform' - AI simply being a computational realization of cognition. This became the theoretical basis the definition of Wipro's HOLMES platform capabilities. In the year 2014, we started developing bots for our IT helpdesk using a corpus of historical information and applying machine learning and natural language processing techniques. By September of the same year, we had deployed 10 bots working in the area of help desk, processing more 10,000 help desk tickets per day. They continue to function taking up responsibilities of categorizing issues, assisting in task assignment and will eventually move to automating resolutions. They learn continuously and are always on 24x7. Clearly now in 2015, it's a race with the bots in the IT services industry - it's not one against the other but about how can humans and machines can collaborate and work together. And this is where, Applied AI, with its capabilities of building bots, takes the center stage. There is possibly another age coming our way in the future, driven primarily by what is called strong AI or natural general intelligence. Strong AI can build machines that think like human beings and have consciousness like AVA from the movie ex-Machina. While we foresee it to take a long time to materialize, the exponential growth of computing technologies could possibly bring AVA amidst us earlier than we expect. In a series of blogs, I will discuss the technologies related to each of the key capabilities in the HOLMES platform - how the platform is built and what it can do. If you have an interest in tracking progress in the AI space follow me on here. The game is a foot! AI, Algorithms, Applications, Artificial Intelligence, Computation, Data Structures, HOLMES, Winter
Computable simply means that there should be algorithm(s) and data structures with some representational techniques that can compute any of the cognitive functions such as believing, consciousness, emotion, language, learning, memory, perception, planning, problem solving, reasoning, representation, categorization, concepts, mental imagery, sensation, thought, etc. So how difficult is it? Is Cognition actually computable? Not all but some aspects of cognition can be computed today using computational linguistics, computing statistics, computational geometry, and computational learning.
Obama announced his 'Computer Sciences for All' plan in his weekly address on Saturday as he emphasized on the need for teaching the subject as a "basic skill" to all children across schools in the country