Artificial Intelligence: The learning curve is moving upward

16, May 2017

AI has been an intrinsic quest of the human race right from the moment we invented the computing machine. The urge to make computing machines more intelligent than ever before, in fact signalled the birth of algorithms and complex application development. Since then, the growth of AI along with its challenges has been constantly changing, similar to the advancements in the IT industry.

Over the years, there have been innumerable ways in which a computer is used to solve problems, that vary based on industry, knowledge and routine steps, and processes. It is no surprise that computers can easily solve problems in a highly matured and process-driven environment as opposed to problems that require knowledge and experience to solve them.

Age of maturity in machine-learning

Computers and software applications have matured to a large extent to solve problems through simulation models (for example, virtual chess). The sheer power of simulation, with various permutations and combinations, has been revolutionizing problem-solving areas. The other is the knowledge and rule model in which basic concepts of database and rules are harnessed to arrive at solutions.

These are certainly effective in a world where there is linear correlation of data, rules and patterns.

The human mind works purely based on acquiring knowledge, making mistakes, gaining experience and consequently – applying rules help us complete the cycle of problem-solving. As far as machines are concerned, how close we can get them to start thinking like human beings is the key to success of AI.

Most systems today requires us to feed content into it, and the rules can be respectively applied. What they do today is ensure that the rules are applied to the content in a fast and efficient way.

Today, learning systems move the pointer closer to the real AI. They can add new content and build new rules. These learning systems (Eg; search engines) are becoming a lot more matured in problem-solving abilities. As we keep using them, the algorithm gets perfected. However, another aspect of human intelligence is the capability to eliminate options. This is yet to get perfected in machine-based learning.

Heuristic models are widely used to arrive at the closest options to solve a particular problem. While these may not be completely accurate, they get us to the expected outcome.

Even though, AI concepts have been around for a long time, recent technology breakthroughs have opened the floodgates for inexpensive AI solutions to hit the market.

Big Data Technologies

Machines and systems are not designed to store unstructured data such as images, sound, video, etc. because of its inherent structured model of storing data which were just texts, numbers and some special data fields. With Big Data, the ability to store all types of information has become feasible. Once stored, we can leverage the power of processing to achieve the objective of getting closer to AI.

Improved algorithms

The human brain has in-built perceptions to recognize patterns in voice, face, touch and other sensory emotions very easily. But the same has been difficult or near impossible for machines. Until now! These days, with the rise of highly-advanced algorithms, machines can leverage natural language processing, voice recognition, image processing, pattern recognition and facial recognition, and help us build sophisticated AI applications.

Need of the hour across industries

The service catalogue is a key element of service delivery and is the most critical and complex part of any customer on-boarding experience. But is there a way that it can be automated and made to self-learn by leveraging the knowledge base that is already available?

Well, if we can co-relate the logs, events, process, and operations in a simple model, we can automate the process of creating service catalogues while integrating them to other toolsets like ITSM, alerts, capacity management, availability management etc.

Another instance is the possibility of automating on-boarding agents in the recruitment process; right from when a resume is received from a candidate to the interview process, selection and joining formalities. If we can automate the entire process, right to the call center floor and workforce optimization, including the agent’s performance metrics through QM, it can substantially save time, money, and resources.

There is no denying that AI is here to stay. And the horizon is wide for machine-learning to evolve into something more sustainable and cutting-edge. The only question that remains is how quickly we adopt it to add true value in the digital era.