Companies Are Testing Whether AI Can Solve Legacy Tech Problems

A lot of companies still depend on COBOL (common business-oriented language), the programming language designed for businesses, which came out over 60 years ago. However, practitioners of this language are from the baby boomer generation and are now retiring, so it’s getting harder to source specialists in this field. To tackle this problem, executives are testing out generative AI tools to catch their IT functions up to modern tech trends, according to The Wall Street Journal.

Generative AI has been rapidly impacting businesses, and technology leaders have been jumping on the bandwagon to streamline and renovate their information-technology systems to get rid of outdated functions used in the past.

Amin Venjara, chief data officer of 75-year-old payroll-processing company ADP, said, “A big problem that we have in our space, and legacy companies have, is we have COBOL running around.” He added that the number of developers who know how to converse in COBOL is dropping significantly.

The use of generative AI is being explored by the Roseland, New Jersey-based company to “translate” its mainframe code from COBOL — which is still commonly used by banks and financial services firms — into Java, which is fairly new to the programming scene.

This transition would reduce the need to scope out and train experts in COBOL, who are hard to find these days, and open up the space for specialists in more modern languages such as Python.

Making this shift can involve software updates and moving to cloud-computing platforms. This has been a key focus for chief information officers, but it has become even more critical now as CIOs are looking to improve efficiency while keeping an eye on the costs of IT.

Over the last year, generative AI-based coding assistants from Amazon, Google, IBM, and GitHub (owned by Microsoft) have come to the scene to aid developers with tasks like auto-completing code snippets and writing code documentation. Some developers have estimated that coding assistants increase productivity by 25%, especially with tasks such as spellcheck and autofill, which help people write documents.

IBM, which still has a huge dependence on its mainframe business and provides support to a broad range of customers that rely on large computers, is marketing its watsonx.ai AI coding assistant to help customers quickly and easily get to the bottom of their problems related to their legacy technology systems.

Skyla Loomis, vice president of IBM’s Z mainframe software, said, “Our clients haven’t always invested as much in their applications over the decades, so that’s where they start to run into risks, skills challenges, knowledge gaps with very expansive applications with tens of millions of lines of code.”