AI is an often overused and misunderstood term. Artificial intelligence is simply a program/system that mimics human-like-intelligence. A simple example of this would be summoning a voice assistant on your phone. The phone listens for a specific phrase, and when it hears it, it launches the program.
Machine learning (ML) is a subset of AI which often causes confusion and misuse of the term. AI will typically give answers based on the “rules” defined by the algorithm. Machine learning is when the AI tries to learn what action it should take based on its data. Hence the learning in machine learning.
Every AI system will need training, and your goals and the data define your option. Supervised learning is when you feed the algorithm known questions and answers. The algorithm will learn based on this information and provide answers for future questions based on the previous data.
AI can be a powerful tool for identifying patterns when there seem to be none. This form of AI is referred to as unsupervised learning and is the practice of feeding huge amounts of data into an algorithm with no answers, questions, or direction. The system will learn all it can and attempt to identify partners and solutions.
Natural Language Processing (NLP) is the technology used to understand and interpret human language. When NLP is used with AI, it allows the two technologies to work in harmony to identify keywords or phrases using NLP and take the appropriate action (rule) or categorize the data source (depending on the form of AI).
AI is one of the hottest topics in IT Service Management. Everyone is looking to add a virtual agent, coach, or other AI form to their systems.
So why has it been so difficult for so many to reap the benefits?
Our guide will present four of the most common areas of contention when it comes to AI and ITSM and gives some tips on how to make your adoption smoother.
Virtual agents or bots are designed to simulate real service desk agents. When discussing bots in the realm of AI, it is important to understand the difference between a bot powered by automation and one powered by AI. A chatbot that has canned responses based on predefined choices does not constitute AI. This type of bot is classified as a rule-based chatbot. For a chatbot to be considered AI-powered, it must have some form of algorithm to set the rules used.
AI-powered agent assistants are often very similar to virtual agents except in one key characteristic; they interact with agents, not end-users. Agent assistants will use algorithms to identify similar incidents or interactions based on a scoring mechanism using natural language processing (NLP) and potentially supervised training. To allow the algorithm to move past basic AI into machine learning (ML), the support persons can score the suggestions based on their usefulness.
Prescriptive maintenance (RxM) is the art of learning when a component should fail and provide solutions to help with decision making. The combination of machine learning (ML) with data collected through sensors for predictive maintenance and internet-of-things (IoT) compatibility will help prescriptive maintenance on its journey to becoming mainstream.
Every human has some routines that they follow in their day-to-day life and even their workplace. AI-enriched user profiling applies machine learning algorithms to identify strange usage patterns such as the sign-on procedure, i.e., time, location, device, browser, or unusual usage, such as deleting large amounts of data during the night.
Today, processes are automated based on rules using workflow automation. Future workflow engines will have workflow activities that make decisions based on AI and machine learning (ML). Complex combinations of If/Then conditions in workflows will be replaced with dynamically adjusted, score-based decisions that continuously learn.
Artificial Intelligence is making its way into all areas of service management. The first versions of AI tools are being implemented, tested, and are evolving.
Learn more about the different AI technologies for service management including:
AI has a wide range of potential use cases, directions, and implementation possibilities. When you start to think about adding AI, it is important to identify which areas you want to improve. At this point, you don't need to consider technologies or vendors, simply the business cases you wish to solve. Don't forget to be critical of your choices because it usually takes thinking outside of the box to get the best benefits from AI compared to your competitors. Learn more about choosing a direction and other tips from our blog "Get your ITSM ready for AI in 5 steps".
AI can be a complicated topic but it is still approachable for every organization. Once you have identified the business cases or general areas you wish to add AI to you should start evaluating your data and abilities. AI requires certain skills surrounding data analysis, and data to work. You can supplement analysis skills by using a consultant, but you can't supplement poor data quality. It is important to test your data to make sure that it is of high enough quality to provide statistically significant results. You can find out more about data quality from our blog "Rate your Data Maturity for AI & 5 Tips for Improvement".
Once you have your data in order and know what type of AI technology you are looking for, it is time to start searching. Many companies specializing in AI and many large software vendors are adding AI capabilities to their platform. You need to consider the simplicity of these systems, integration capabilities, and security. How easy will they be to add, and where will your data be stored. All of these are topics that you need to evaluate especially as privacy and data protection legislation expands. Find out more about our AI implementation from our blog "Why a Virtual Coach? – The Inside Story".