Artificial Intelligence and Machine Learning Are Coming Up on the Mainframe’s Horizon

Emerging technologies like artificial intelligence (AI), machine learning (ML), and deep learning (DL) are rapidly developing and evolving—as is what we can do with those technologies.

According to Jeff Bisti, Cognitive AI Architect at IBM, the mainframe is on the periphery of being able to participate in machine learning; in the future, IBM Z® will play an even more important role in the technology’s development.

The Mainframe Opportunity

Bisti explained that mainframe will be able to help with some of the biggest problems within AI: getting access to quality information. Data sets are often months or years old, resulting in inaccurate models.

That’s where mainframe comes into the picture. One thing almost every mainframe shop has in common is that they wield the power of having a lot of data, and much of it being real-time data.

“Mainframes are responsible for many of the world’s biggest businesses,” Bisti said. “There are so many insights that can be drawn from the type of day-to-day data we have. It’s time we take that data and make it start working for us.”

Even better, mainframes will be able to provide data in a way that makes the job of a data scientist a lot easier. In the distributed world, data is also distributed, so to speak. Data scientists may have to go through different processes to retrieve data, and data from different sources may be in different formats or languages. On z/OS, however, there are products like IBM Open Data Analytics for z/OS (IzODA), for example, that make it possible to present normalized data.

Providing data from the mainframe is also a much more secure alternative than simply zipping up databases and sending them around, Bisti explained.

“Most company security compromises come not from hackers breaking the master password, but from someone leaving valuable pieces of data around on the cloud or on a USB thumb drive where anyone can get it,” he said. “In the push to do AI, people sometimes aren't the most careful with the data they're using. Data control and pervasive encryption is one of the big pillars for the mainframe.”

A Cultural Challenge

One of the biggest challenges at the intersection of mainframe and AI will be cultural, rather than technical. Bisti suggested that the rapid nature of evolution in this field may take some adjustment on the part of mainframers.

“There are people doing amazing things with machine learning, but everything we’re doing will look so outdated, so quickly,” Bisti explained. “We have to get used to things advancing and being reinvented on a much shorter cycle than we’re accustomed to. There won’t be a piece of technology in this space that you can hang onto for more than say, three years. On the mainframe, this will be a challenge. People will have to be constantly learning to reinvent themselves in this constantly churning ecosystem.”

How Can Mainframers Get Started?

Whether you’re intrigued or still a bit skeptical, it’s going to be incredibly helpful to actually see these technologies in action. The good news is, people are already using AI/ML/DL to solve problems. Bisti recommends checking out Kaggle, a hub for big data and analytics.

Companies or individuals upload anonymized data sets to Kaggle, such as monthly sales or data on Kickstarter projects, and often start competitions for the best algorithm to solve a specific problem. The rewards range from money to job offers, but the really interesting part is that many people show their work, so newcomers can start to understand how people are using the technology. With thousands of companies and data sets, Kaggle is a very active community of people doing work with AI/ML/DL right now. According to Bisti, it’s getting better and more intuitive, and now is the time to start playing around.

Jeff will be speaking more about these topics in a session about practical AI at SHARE St. Louis. He hopes to give mainframers the tools they’ll need to be successful in the future.

“I want to break down what’s actually happening in AI/ML/DL so people can see just how empowering it can be,” he explained. “Session attendees will be able to see how the pieces fit together, so when these products and offerings start coming into their shops, they will be able to make more informed decisions and select the right software and solutions that will help their business. They won’t leave my session a data scientist, but they will know enough to figure out the next step, should they choose to pursue AI/ML/DL.”



Be sure to join us at SHARE St. Louis, August 12-17, for even more on artificial intelligence and machine learning. Jeff will host “Practical AI: Paranoia Aside, What Can AI/ML/DL Do for Me?” Thursday, August 16 from 3:15 pm to 4:15 pm – log in here to save his course to your SHARE profile.

Recent Stories
Using IPCS to Solve CICS Failures

Tips and Tricks: The Transition to Remote Work

Master the Mainframe 2005: Anna Clayton's Springboard to Success