Generative AI involves mainframe software modernization with a mannequin skilled on greater than 80 code languages and 1.5 trillion tokens of knowledge.
IBM COBOL-to-Java translation. Image: IBM
IBM introduced in the present day watsonx Code Assistant for Z, a generative AI-assisted answer for COBOL-to-Java mainframe software modernization. It opens up new use circumstances for watsonx Code Assistant, specifically transferring and validating COBOL purposes on IBM Z.
Watsonx Code Assistant for Z is anticipated to be out there globally as a service within the fourth quarter of 2023. IBM will exhibit watsonx Code Assistant for Z in September at IBM TechXchange in Las Vegas.
Jump to:
Make code migration from COBOL to Java simpler
Mainframe computing equivalent to IBM Z has been a spine of IBM’s enterprise for many years. The IBM Institute for Business Value discovered that organizations are 12 occasions extra seemingly to make use of present mainframe property fairly than improve their software estates to utterly new environments within the subsequent two years.
More must-read AI protection
Watsonx Code Assistant for Z is meant to make it simpler to improve whereas utilizing present mainframe property, with generative AI standing in for potential talent gaps or strained assets. In many situations of software modernization, organizations transfer all software code to Java or a public cloud; IBM desires to verify prospects don’t lose sight of the unique cause they selected IBM Z as a substitute of a public cloud.
Watsonx Code Assistant for Z is supposed to be used in software modernization, app improvement, giant libraries and information retrieval; specifically, it’s designed to make translations from COBOL to Java sooner. Other instruments that mechanically change COBOL purposes to Java exist, however they will produce garbled code that’s exhausting to take care of for builders used to working in Java.
The COBOL information processing language is comparatively straightforward to make use of, but additionally more and more hardly ever used, and it’s a procedural language as a substitute of an object-oriented one. The object-oriented Java has broader performance, and its complexity will be useful. Watsonx Code Assistant for Z can refactor, remodel and validate COBOL code throughout software modernization on IBM Z (Figure A).
Figure A
A diagram of how Code Assistant for IBM Z suits into the modernization life cycle. Image: IBM
Today, purposes are sometimes “tangled-up monoliths” that make code migration time-consuming, sophisticated and dangerous, stated Skyla Loomis, vice chairman of IBM Z Software, in a pre-briefing for press on August 17.
Enhance the transition to generative AI
Specifically, Code Assistant for Z makes use of the IBM watsonx.ai basis mannequin to rewrite, assess, replace, validate and check code. The generative code mannequin at the moment accommodates 20 billion parameters and was skilled on greater than 80 code languages and 1.5 trillion tokens of knowledge. IBM claims its Java translation outperformed ChatGPT 88% to 32%.
Some performance will come from IBM’s Application Discovery and Delivery Intelligence stock and evaluation instrument. After utilizing the Application Discovery and Delivery Intelligence instrument, prospects can use watsonx Code Assistant for Z to:
Refactor enterprise providers written in COBOL.
Transform COBOL code to Java code with an eye fixed towards optimized design.
Validate the ensuing Java code utilizing automated testing.
One use case is perhaps delivering Ansible Lightspeed — a generative AI service designed to make it simpler to ship Red Hat’s automation instrument Ansible — with watsonx Code Assistant, which creates Ansible-tuned fashions.
How watsonx Code Assistant for Z integrates with VSCode
During the press demonstration, IBM Fellow and CTO for Z Software Kyle Charlet demonstrated extracting COBOL code from an insurance coverage software for refactoring. Watsonx Code Assistant for Z may hint code by related information buildings, extracting the precise code paths wanted for that individual software versus the others utilized in related duties.
SEE: IBM is betting on Meta’s Llama 2 within the race to safe territory in the way forward for generative AI (TechRepublic)
From there, one may export the code paths to Visual Studio Code, inside which watsonx Code Assistant for Z may present suggestions and suggestions.
Watsonx Code Assistant for Z additionally evaluates that the COBOL code and the Java translation are semantically equal and have the identical end result.
AI mannequin will likely be skilled on open supply or attributed code
The AI mannequin utilized in watsonx Code Assistant for Z was initially skilled on CodeWeb and is now being tuned on enterprise Z COBOL and COBOL-Java pairs, identified Charlet. He famous that the AI mannequin works beneath an open supply license and wouldn’t be straight copying anybody’s authentic code with out their permission.
“Attribution of code will be a key priority. Contributors will know if their code is used to train that model, and they can opt out of that experience,” Charlet stated.
Watsonx Code Assistant for Z’s place throughout the bigger IBM AI ecosystem
Watsonx Code Assistant for Z is likely one of the many branches of IBM’s efforts so as to add generative AI to its merchandise. It sits throughout the identical AI and information platform class as a number of different elements of watsonx, together with:
watsonx.ai, which trains, validates and deploys machine studying AI and basis fashions for generative AI.
watsonx.information, which permits enterprises to scale AI workloads utilizing their information with a fit-for-purpose information lakehouse for AI workloads.
watsonx.governance, which supplies information and AI governance for accountable, clear and explainable workflows.
“Our focus is on a full stack of solutions starting all the way down at the infrastructure which of course includes IBM Z and IBM cloud and goes up to our Center of Excellence for consulting and Client Engineering,” stated Keri Olsen, vice chairman, IBM IT Automation.
“By bringing generative AI capabilities through watsonx to new use cases, we plan to drive real progress for our clients,” stated Kareem Yusuf, senior vice chairman, product administration and progress, IBM Software.