The two main areas of focus where AI has been showing itself in a IT Service Management company is classification based machine learning, as well as virtual agents. While many other opportunities and use cases exist, these two have manifested themselves in many products on the market today.
For machine learning, having a human identify the valid training data is key. In rank & retrieve style algorithms, as those commonly used in recommending related results to users, doing manual labeling helps the performance greatly. In ITSM, technicians can manually label valid question and answer pairs on issue resolution. This helps the system recommend the proper answers in the future.
For virtual agents, humans are needed to train NLP engines on multiple utterances of how a given intent can be phrased. In addition, humans need to decide the conversation tree that bot will follow. This creates a bot that can augment their efforts in providing fast service and answers to questions. Complexity will be a big factor in dividing the work between people and machines. Low complexity, repetitive work that can be taught to the machine will be handled primarily by machines, while new and complex problems will remain in the hands of people.
In this regard, more value is given to higher skilled labor, and lower skilled workers may see their positions reduce. The reduction of human labor required is a very clear side effect of the computationally superior abilities of machine learning in classification settings, as well as the ability for virtual agents to augment or replace tier 1 technicians. Management is increasingly being held accountable to ensure that tasks or processes they manage utilize these latest AI trends.
- Machine learning will help reduce human error in classification, and save time for existing technicians. This may not always directly reduce labor, but will rather improve the experience and reduce the time required to conduct classification tasks. New jobs will be required to train and maintain these models.
- Virtual Agents can generally reduce the number of tier one technicians by 20%, with some companies striving to reduce tier 1 technicians to 0 by 2020. The concern is about the reduction of low-skilled workforce, so it will be critical to see how new roles or training can evolve to allow technicians to remain at a company. This ‘upskilling’ will be key. New jobs will be created to maintain, train, and design the conversations and NLP abilities of the virtual agents, however this will be far less than the amount of technicians reduced.
To summarize the concerns, there are two primary areas for consideration as organizations adopt AI technology in ITSM:
- Privacy Concern: AI in general is built on data. When looking at auto summarization and speech to text services to summarize customer responses and interactions, there is concern over privacy and over IP generated. Organizations must be explicit on data collected, and in addition be sure to use anonymization and aggregation techniques.
- Workforce Reduction Concern: Given the scalable nature of AI solutions, and their ability to reduce manual labor from repetitive or process driven tasks. Organizations must focus on workforce training to enable lower skilled workers to begin solving higher complexity tasks. They must also prepare new roles to maintain the solutions.