Artificial Intelligence (AI) aims to empower computers, or computer-controlled robots, to perform tasks that are usually performed by humans, imitating the human ability to solve problems and make decisions. This goal is now achievable—in part—thanks to a combination of hardware and software that have allowed for the development of machines equipped with autonomous learning and adaptation capabilities that are inspired by human learning models.
These tools, especially since ChatGPT was unveiled just over a year ago, are already having a significant impact on how business is conducted around the world, completing tasks with a speed and efficiency that would not be possible for humans. At present there are three types of AI, classified according to the extent that they are able to imitate human characteristics, the technology they use to do so, their real-world applications, and theory of mind:
Narrow Artificial Intelligence (ANI), which has a specific range of abilities.
Artificial General Intelligence (AGI), which matches human abilities.
Artificial Super-Intelligence (ASI), which exceeds human abilities.
ANI is the one that interests the market the most. It’s already available and ready to spawn a number of useful products that complement, rather than substitute, human activity.
As it happens, ANI is the only type of artificial intelligence that has been successfully developed and adopted to date. It’s main characteristic is that it is designed to perform single tasks, or, if you prefer, target single goals. While ANI characteristics may appear intelligent, they are bound by constraints that prevents them from replicating, or even imitating, human intelligence. Rather, ANI simply simulates human behavior within a narrow set of parameters and within a well-defined context. As an example, ANI is what powers self-driving cars. It’s what allows Amazon.com to recommend a book based on your purchasing and reviews history and what allows Alexa to respond to play a song you like. ANI powers ChatGPT, which as smart as it may appear is still designed to perform a well-defined task. ANI is also the technology that allows IBM watsonX to work which performs the invisible—to the larger public—but highly useful task of perpetuating the use of COBOL, facilitating more cost-effective and reliable data migration. It sounds simple and rudimentary; but it’s an important achievement.
Announced in October 2023, IBM’s watsonX Code Assistant for Z is a cloud service, which uses generative artificial intelligence to accelerate code generation and increase developer productivity. The watsonX Code Assistant family can understand up to115 programming languages acquired from learning 1.5 trillion tokens with 20 billion parameters. The scale of this project is extremely ambitious, and IBM intends to transform watsonX into one of the largest generative AI foundation models for automating code writing. IBM has literally developed a tool that rewrites code using artificial intelligence: watsonX Code Assistant for Z combines IBM’s artificial intelligence expertise—with IBM subsidiary Red Hat’s strength in IT automation. Red Hat Ansible is an open-source automation platform that can automate IT operations, from configuration to deployment, and resource management. By integrating watsonX Code Assistant with Red Hat Ansible, developers can write Ansible playbooks with AI-generated recommendations, reducing errors and increasing developer productivity. Additionally, the code generated conforms to industry standards, ensuring greater quality and reliability. It seems that IBM’s overarching goal is to create a basic model that can be infinitely applied to all existing languages, revolutionizing the rewriting times of business applications—competing with Microsoft, Amazon, and Google generative AI products.
WatsonX addresses a fundamental issue: programmers often spend 90% of their time revising and correcting existing code. Until recently, programmers have had to be experts in both the latest language innovations, and in those used decades ago—as frequently happens in mainframe computers.
Significantly, for large institutions with huge historical data to process, watsonX allows for a rationalization of costs while expanding the pool of possible developers to deploy (or hire) for a given task, reducing the training time of new recruits; because it sustains the viability of COBOL. It does so by using the same generative AI principles behind the aforementioned applications (i.e. ChatGPT) to provide code recommendations based on natural language input or existing source code. For example, a developer could enter a natural language request like “Create a function that calculates debt to income ratio or compound interest.” IBM watsonX Code Assistant for Z would then generate the corresponding code in Java. This not only saves time and effort but also reduces errors, increasing productivity.
COBOL (Common Business-Oriented Language) has been widely used in business applications for decades but has become less common due to the emergence of new languages and technologies. Nevertheless, COBOL is difficult to give up. It was designed to be easy to read and write, presenting a syntax similar to English and, in general, close to natural language. For this reason, many businesses have retained COBOL to develop data processing systems, manage financial calculations and for many other business applications. However, integrating modern features like mobile access or real-time data analysis into COBOL-based systems can be challenging. Thus, watsonX Code Assistant for Z represents a crucial step in modernizing enterprise applications, especially those based on COBOL, a programming language that has played a critical role in many vital business and operational processes globally. According to a 2022 survey, over 800 billion COBOL programming lines exist worldwide, yet COBOL development experts are becoming ever scarcer; and migrating COBOL code to Java is costly and demanding—the more so for the kinds of projects needed by larger enterprises.
Another powerful aspect of watsonX Code Assistant for Z is the ability to customize base templates with company standards and industry best practices. For instance, a bank could customize the template to ensure that the generated code meets financial industry regulatory and compliance specifications, aligning the solution with the specific needs of the organization. This alignment ensures that the solution is not only technologically advanced but also strategically aligned with business objectives.
IBM watsonX Code Assistant for Z symbolizes a new era in application modernization and software development. Its introduction marks a turning point that goes beyond simple efficiency and productivity, touching on fundamental aspects of technological innovation and business agility.