The Next Generation of Analytics with Automation is AI 

In the IT industry, we often hear words such as analytics, automation, artificial intelligence (AI), machine learning (ML), predictive and prescriptive. But we don’t always pause to consider what those words actually mean—to the point where, in some cases, the terms (like AI and machine learning, for instance) start to get used interchangeably.  

For many years, IBM Z has had analytics with automation.  We’ve used SMF, log, and system data to analyze and manage the system, automating the repetitive, manual processes that humans were doing.  Fast forward to today, and the introduction of technologies such as the Telum processor gives Z customers next-level analytics with automation.  It brings the capability to leverage all that data - right on the platform that’s known for its speed, scalability, and reliability – to drive better insights and automation that can power the business forward.  It’s like taking the leap from driving to flying – it’s a faster, better, smarter way to make the journey.  The first step is understanding where to go and why you’re going there.   

Differentiating AI From Machine Learning 

Eberhard Hechler, Executive Architect at the IBM Germany R&D Lab, discussed these topics at the Guide Share Europe event titled “On the Wave of AI and Beyond.” His presentation, “AI and Cognitive Analytics—When Will Robotics Dominate the World?” explored the degree to which autonomous and self-directed learning by machines is possible.  

Hechler explained that machine learning is the process of layering algorithms on top of each other to build models and made the distinction between Artificial Intelligence and Machine Learning. In the wider context, Artificial Intelligence is “intelligence demonstrated by machines that mimic cognitive functions that humans associate with other human minds, such as learning and problem solving.”  Machine learning is one subset of AI techniques that includes rules-based systems, optimization, prediction, and prescription.  What businesses are ultimately looking for from AI is a way to automate decisions that then automate actions to accelerate their business.  

Analytics and Automation Cannot Work Alone 

However, before there can be AI or machine learning, there must be data to analyze, and business is beginning to understand the power of the data it’s been collecting for many years. In “Gartner Top 10 Data and Analytics Trends for 2021,” for example, the analyst notes, “Business leaders are beginning to understand the importance of using data and analytics to accelerate digital business initiatives. Instead of being a secondary focus—completed by a separate team—data and analytics is shifting to a core function.” 

Analytics alone is simply reporting. It’s the evaluation and correlation of a massive amount of data, but by itself, it’s simply a report. It still must be read by a human, who determines the next action, then that action must be taken. While the information is useful, the enterprise doesn’t save time and increase efficiency on that data alone. The organization must be able to derive insights, make decisions, and act on the information to truly make a difference.  

Somewhat similarly, automation alone requires people with deep expertise and legacy knowledge to develop the rules to automate processes properly. Without insights from data, rules are developed in a vacuum – automation requires context.  It needs to understand, from data, the right confluence of conditions to trigger automation, and the appropriate action to take.  To make matters more difficult, as business leaders and IT department managers around the world know, people with the subject matter expertise to skillfully craft automation are rare, or they are simply overwhelmed with other work.  

Analytics + Automation = Fit for Business Purpose 

It’s when you connect analytics and automation that infrastructure becomes more efficient and Fit for Business purpose. Automating tasks through analytics can reduce response times, enforce consistency, and reduce errors associated with manual intervention.  

The business advantages of connecting analytics and automation are numerous:  

• The skills gap problem is somewhat reduced as more tasks become automated 
• The data that has been collected yields business benefits  
• Staff can spend time performing other, more complex tasks  
• Fewer errors and more consistency are a result 
• Overall service is improved, increasing customer satisfaction and your edge over the competition 

As I noted earlier, IBM Z has been doing this for years. It’s one of the reasons I say that Z is fit for business. However, there’s more. When you add modeling and machine learning, the next steps are towards predictive, prescriptive, and real AI technology.  

Data Models, Prescriptive Analytics and Predictive Technology 

According to ‘Dealing with the AI Data Dilemma’, a paper from the IBM Institute for Business Value, the key to being successful in any AI project is to first understand why the business needs to do AI, rather than how to do AI.  “Without sufficient understanding and regard for the larger business issues—why the company is doing AI—proofs of concepts and research-type projects can proliferate without benefiting the business.” Teams first need to ask two basic questions: “What business problem are we trying to solve? And how should we best solve it?” Sometimes, the most sophisticated AI may not be the best answer. 

This Needs Matching decision guide, from the same paper, provides a pragmatic approach for making decisions about your journey into analytics and AI.  

Once you’ve made those decisions about the level of analytics appropriate to meet the business needs, determine your data sources and collect the data.  Once enough information has been accumulated and analyzed, it’s possible to begin building models and making predictions based on those models.  

For many years, the capabilities of Descriptive Analytics have provided reporting to show where we’ve been and where we are today.  For example, a retailer with a few years’ worth of data could generate reports to show how much CPU capacity they’d used.  Over time, forecasting algorithms came about for capacity planning – Predictive Analytics to project when they would likely need more CPU capacity to handle business growth.  

The next level of sophistication is Prescriptive Analytics. These models predict the need for additional CPU capacity and the analysis prescribes how much and when to add it. The entire process is simplified, more accurate, and less stressful than manually adjusting to add CPU during each busy season.  This is the first half of the AI equation noted earlier - automating the decision by using the data.   

The next step to “automate the decision and then automate the action” is for AI predicting when capacity will be needed, prescribing how much to add, and then taking automated action to add it. 

“Intelligent automation is a new capacity that enables processes to perform in ways that optimize the amount of human support needed,” according to “The Evolution of Process Automation,” an IBM Institute for Business Value report. By implementing technology to handle tasks that happen too fast for humans to respond or that happen so often humans spend too much time on them, the enterprise gains something that cannot be overvalued: the strategic and creative thinking of humans.  

What this evolution means for operational AI in the IBM Z ecosystem is that we can have the hardware and software continue to expand on what it does best – reliably process, analyze, and learn from vast amounts of data at phenomenal speeds, and act on that information with artificial intelligence, ideally freeing people to use the unique abilities of human intelligence that AI hasn’t yet mastered: intuition, creativity, innovation, and reasoning.