There’s no doubt that the future of IT service management (ITSM) is very much entwined with the opportunities of artificial intelligence (AI) and machine learning in particular over the next few years. But what does this statement mean in operational terms and what do organizations need to be doing to ensure that they use AI technology to improve their capabilities?
To help, this article seeks to answer some of the key questions that you and your colleagues might have related to AI technologies and ITSM.This article by @ITSMNinja seeks to answer some of the key questions that you and your colleagues might have related to AI, machine learning, and ITSM. #AI #machinelearning #ITSM Click To Tweet
The Perpetual Challenge When Introducing New Technology
One of the challenges with any technology is that we get enamored with the technology and everything it can do and forget the basics of any project that releases a new service: what is the purpose? How will we know if we succeeded? How will it impact (increase value) for our stakeholders? Understanding what is needed to support and improve AI is also critical to any initiative’s ongoing success. How will the use of AI technology align with your organization’s values, goals, objectives? Just because you can do something with AI doesn’t mean that you should.
People often think AI technology initiatives will understand their environment and uniqueness, forgetting that they continue evolving and changing. Not only are clear objectives necessary, but so too is an understanding of how organizational and technical knowledge will be captured, structured, reused, and improved. How will all of the implicit knowledge in an organization become explicit so that the machine learning technology can reuse it to improve the stakeholder experience? Where is the quality assurance that assures changes conform to governance directives and controls?Just because you can do something with AI doesn't mean that you should – @ITSMNinja #AI #ITSM Click To Tweet
The Opportunity for AI for ITSM Organizations
Today, there are three significant opportunities with AI for ITSM organizations. The first is in providing a more interactive response, e.g., chatbots or virtual assistance, referred to as Level or Tier 0.5. It’s self-service without the customer needing to know how to format the question or use the specific terminology (at least that’s the concept). A good knowledge management practice can also support multiple languages using the technology, a significant advantage for organizations who need to support numerous languages. The challenge, however, is that today people translate all the statements made and can focus on the myriad of terms used for what appears to be a simple question.
The second area is using the AI technology in assisting support representatives with more complex issues. Most people won’t remember everything, especially in an environment where the rate of change is high or there is an infrequency of answering questions on a particular topic.
The third area is the use of machine learning in analytics to identify correlations that humans cannot do cost-effectively. Service management systems have a lot of data, but little of it is analyzed or used to make decisions. Not only can AI technologies correlate and identify causes, but it can also predict/forecast performance based on historical data. Since people are creatures of habit, this historical data can provide the basis for accurate forecasts, facilitating better decisions and better risk management.There are three significant opportunities with AI for #ITSM organizations says @ITSMNinja. Here he explores. #artificialintelligence Click To Tweet
AI and 5G
One technology that will facilitate AI adoption and accelerate the value of these AI-driven services is 5G. Not only has computing power been a factor in the usability and acceptability of AI-driven systems, but network speed has been a factor too. The use of machine learning requires a significant amount of data, which often needs to be real-time, which can only be met with 5G speeds. This is also true of the Internet of Things (IoT), the embedded devices that exist in many places, but are often not effective due to the latency in networks currently. With the help of 5G, AI-driven robotics acceptance and use will grow significantly in the next few years.
People and AI
One unanswered question is, should the users know that they interact with AI technology, not a person? Ultimately, the user is looking for the easiest path for a resolution, but should they understand how their data will be reused? Should they know that they’re not interacting with a person? How do you ensure no dead ends or endless loops which will quickly frustrate users, just as a 404 or “page not found” error does today for self-service?
The core of an effective AI-enabled service is the knowledge base, which needs to be quickly (real-time) updated to ensure correct responses are available. This type of knowledge practice is required for machine learning technology to find the answers and edit/update them based on what is learned. So, where is the initial data you’re using coming from? What is the opportunity for bias in that data? E.g., if you think 80% of the incoming issues to the service desk are network issues, how will that bias impact your escalations for issues that it doesn’t know the answers to?Here @ITSMNinja queries: should the users know that they interact with #AI technology, not a person? What do YOU think? #machinelearning #ITSM Click To Tweet
AI and Bias
The initial thinking was that AI and machine learning would reduce bias; however, what was found is that, without proper training and analysis, vulnerable groups could have their rights impacted. Whenever there is a choice of two actions, there is the opportunity for a bias of some non-fundamental variables. AI-based systems learn to make decisions based on their training (data), including historical or social inequities, even when variables such as gender, race, or sexual orientation are removed. With training algorithms, they’ll mimic how the people conducting the activities do things and will continue to follow these patterns unless it’s changed. But bias detection algorithms need to be part of any system that is self-learning.
Another critical challenge is when machine learning algorithms trained using data in one context are migrated to a different context with different rules about the types of attributes that can be considered. Ethical and legal issues can result without anyone knowing until it’s too late without the proper controls.There's a lot for organizations to learn about how to be successful with AI and machine learning. Here @ITSMNinja discusses some of the critical success factors. #AI #machinelearning #ITSM Click To Tweet
There is a lot for organizations to learn about how to be successful with AI technologies. Some of the critical success factors include:
- Do your customers know that they’re conversing with an AI bot?
- Is your organization transparent with its stakeholders?
- Do stakeholders understand how their data is used?
- Who is checking for bias?
- Do you have ethical guidelines for your AI practice?
- Are the bots and virtual agents robust, i.e., no dead ends?
- How do you confirm and verify that people are treated equitably?
- Are the outcomes consistent and reliable?
- Can stakeholders opt-out from sharing their data?
- What is the level of risk with the technology?
- Do you have the data scientist resources needed to leverage and optimize your technology and human investing in AI?
Hopefully, my article has helped your thinking re how to effectively utilize AI to support your ITSM ecosystem. If you have any comments or questions, please let me know in the comments.