Pseudocode: The next big programming Language

Pseudocode: The next big programming Language

Jonas Hultenius, Software Architecht at Sogeti and one of our global experts in SogetiLabs dives into the topic pseudocode, the new contender that is emerging and challenging the traditional norms of coding. Read his blog here to find out more.

In the ever-evolving landscape of programming, where programming languages rise and fall in popularity, a new contender is emerging, challenging the traditional norms of coding. English, French, German, Hindi and even Swedish (or any other human language for that matter) are now viable contenders in this brave new world of generative AI and large language models. 

While many argue that this is the end of an era and that natural languages will emerge as the winner, I believe that this is only partially right.

To me, there is another contender, and while it is not a programming language in the conventional sense, it is a representation of logic and algorithms that transcends the barriers of syntax and bridges the gaps that natural language (not the AI models) will have a hard time covering. 

Enter pseudocode, not as a supplement but as a language in itself, now redefined by the capabilities of Artificial Intelligence and Large Language Models to become something much more than it has ever been.

Historically, pseudocode has been a tool used by programmers to plan and design algorithms before translating them into a specific programming language. It serves as an intermediate step between human-understandable logic and the strict syntax of programming languages. However, with the advent of AI, pseudocode is set to undergo a transformation.

Imagine being able to describe the logic of a program in natural language, devoid of the intricacies of syntax. This is the promise that AI, particularly LLMs like GPTs, bring to the table. These models have been trained on vast datasets, encompassing diverse programming languages and paradigms, enabling them to understand and generate code-like structures. 

One of the primary challenges in programming is translating the intent of a developer into precise syntax. Programming languages have rules and structures that must be followed meticulously, leaving little room for ambiguity. However, AI models have demonstrated the ability to understand the intent behind human language, allowing developers to express their ideas in a more natural and intuitive manner. 

For instance, a developer could describe a sorting algorithm in plain English, and the AI model could generate the corresponding code, taking care of syntax details. This shift from explicit instructions to conveying intent opens up programming to a broader audience, including those who may not be well-versed in the intricacies of specific languages. 

Up to this point in this logical exercise, we have been able to achieve a lot with just natural language. But it is here that pseudocode would make its triumphant entrance. While a language model can produce code, that code has to be readable to the end-user and logical, and this is the gap that could be filled by pseudocode. Today’s programmers would be able to adopt their understanding of the development process and the core logic and be able to interact with the generative AI team member in producing language-agnostic solutions. 

The pseudocode could then be used by the language model as a prompting language, and the result would be a solution that could be evolving over time, changing both platform and core language as the need arises. 

In this new paradigm, programming becomes less about memorizing syntax and more about expressing creative problem-solving. Developers can focus on the logic, algorithms, and the overall architecture of a solution, leaving the minutiae of syntax to the AI.  

Some developers could choose to live in language-agnostic bliss as programming logic generalists while others can keep their syntax skills and continue as syntactical specialists. The options are endless. 

Lowering the bar of entry could also be an important step in democratizing programming and software development as a whole. 

The traditional path to becoming a proficient programmer often involves learning the syntax and rules of one or more languages. This learning curve can be a barrier for individuals who have brilliant ideas but lack the time or resources to delve deeply into programming and the myriad of languages that exist today. AI-powered pseudocode has the potential to help people to participate in the creation of software and share their ideas and concepts with the world. 

While the prospect of using pseudocode is exciting, it comes with its set of challenges. The main hindrance in the adoption is the gap between high-level logic and low-level syntax. While AI can assist in generating code snippets based on natural language descriptions, there is still a need to bridge this gap seamlessly. Tools that provide a smooth transition from pseudocode to production-ready code will play a pivotal role in the acceptance of this new programming paradigm. 

As AI continues to advance, the future of pseudocode appears promising. This evolution may reshape how we approach programming education, software development, and collaboration within the tech community.   

My prediction is that we will see an upcoming renaissance for good old pseudocode, propelled by the capabilities of AI and the amazing powers of modern LLMs. This shift has the potential to redefine the programming landscape, making it more accessible, intuitive, and inclusive. While challenges remain, the vision of a programming world where expressing ideas is as important as adhering to syntax rules is closer than ever before.  

As we stand at the cusp of this transformation, the next big programming language may not be a language at all but rather a bridge between human creativity and machine execution. 

  • Jonas Hultenius
    Jonas Hultenius
    Software Architect & SogetiLabs Fellow, Sogeti Sweden
    070-518 66 25

    Sorry, this content can only be visible if Functional Cookies are accepted. Please go to the Cookie Settings and change your preferences.