One for the age-old immortal computer demons still among us: COBOLone of the first programming languages to gain favor but was long succeeded by more efficient and easier to understand languages such as Pythonit may finally be on its way.
What is that? Can’t see a white COBOL code anymore? Well, basically you do: In fact, 800 billion lines. Oddly enough, that figure has actually increased from 2017, when it was 220 billion.
At the same time, the number of COBOL experts is decreasing, because COBOL is old, like most companies today. So it’s a guarantee that many moons have passed since the only sysadmin who could untangle all that code, who is definitely older than me, and probably older than you, left the building, or even this mortal spiral.
COBOL issues
If TechCrunch have discovered, companies looking to exit COBOL therefore face two major problems.
First, COBOL experts have rare and in-demand skills, so there is a cost associated with their services (often $100 USD per hour). Second, the sheer amount of COBOL code that still exists means that translating that code will be a time-consuming process. The Commonwealth Bank of Australia found this out when it spent five years and $700 million on this process.
In short, the insurmountable problem facing companies around the world today is all due to a lack of foresight on the part of their predecessors.
So what’s the solution?
Enter computer giant IBM and its Code assistant for IBM Z, which will be previewed in early September 2023, on the occasion of the company’s TechXchange conference in Las Vegas. The idea is simple: if human-solving the problem just isn’t practical, let’s put today’s generative AI on the case instead.
There are some tantalizing promises in the new AI tool – it claims to be able to convert COBOL to Java without sacrificing performance and security (which, until we see what it can do, is just PR bawl), and there’s even room for expansion, such as the generative model that in use, CodeNet, supposedly understands about 80 other programming languages.
IBM isn’t the first company to recognize the need to convert COBOL into readable modern code, or that automation is a novelty, and perhaps the necessary route to get there. But tackling it, says IBM Research chief scientist Ruchir Puri, is not about neutering the more powerful aspects of COBOL.
“If the ‘understand’ and ‘refactor’ capabilities of the system recommend that a particular subservice of the application should remain in COBOL, it will stay that way and the other subservices will be converted to Java,” Puri said while giving an interview told TechCrunch.
All this comes with the usual caveat, though: generative AI is a technology in the making, not perfect, and should not be deployed without human oversight. It is to Puri’s credit that he at least recognizes this.
“As with any AI system, there may be unique usage patterns of an enterprise’s COBOL application that may not yet have mastered Code Assistant for IBM Z,” he says. “It is essential that the code is scanned with state-of-the-art vulnerability scanners to ensure the security of the code.”