What is true Artificial Intelligence (AI), and how can it transform IT service management?

Automation technology is being deployed more readily in IT and business operations, a trend that shows no sign of slowing. An IBM survey found that 80% of companies are already using or planning to employ automation software in the next 12 months. One-in-three companies were influenced by the pandemic to accelerate automation deployment specifically to boost productivity, and improve resiliency through new applications, like automating IT service tickets.

For IT service management (ITSM) leaders, discovering the right intelligent solution for their use case can be daunting. This may be because common terminology (and hype-filled case studies) can obscure, rather than illuminate the underlying technologies and their true relevance. For example, machine learning (ML) is often mentioned in shorthand for AI technology, but the approaches to ML and true AI technologies are different enough to require separate categorization.

What is the difference between ML and true AI?

ML needs no introduction at this point. Its ubiquity has redefined our daily lives, from personalized Google search results to “you may also like…” recommendations on Amazon. But it’s worth revisiting how this technology functions to understand where ML ends and true AI begins.

Statistical machine learning is the process of computers systems using algorithms and models to make predictions or inferences drawn from large datasets, without human interaction. ML models can be supervised—trained on past data to make predictions—or unsupervised—drawing connections and conclusions without past datasets.

ML models have to be evaluated, scored, and trained continuously in order to deliver business value. CIOs who have undertaken ML projects have likely discovered the significant amount of data scientist input and effort involved in producing and managing a useful model. Over time, an AI operations practice may help an organization manage ML models without human intervention, but this is still very much an emergent and edge case.

In ITSM, ML has been useful in trawling large volumes of data to improve user experience, using predictive intelligence to understand user behavior and guide them more seamlessly to resolution through relevant article recommendations, automated search results, chatbots, and more.

Now, true AI is going further, using a suite of AI technologies to predict, contextualize, and respond to information in a system. True AI represents a higher set of cognitive capabilities that make and act on predictions with a high degree of accuracy, improving over time without human-directed training. The distinction between true AI and ML is like having Siri order takeout from your favorite restaurant for you rather than returning a list of top delivery services. It has the potential to create results automatically.

What makes true AI possible?