top of page

AI Will Change DevOps


Technology is eating the world! It is all around us and has become so pervasive over the past two decades that it is literally everywhere and touches everything in modern life. From toasters to cars to watches to glasses and televisions. The adoption and influence of tech is undeniable. And while 20 years is a very short period of time for a new set of tools to become broadly adopted and fully integrated into society, we have somehow reached a new tipping point that will so massively accelerate this trend that the world will most likely look unrecognizable in the next 2 years. That tipping point is the birth of a highly impactful, yet still nascent area of development: artificial intelligence (AI). From general purpose generative artificial intelligent (GPGAI) that can answer almost any question with unprecedented speed and accuracy, to voice and text and image processing chatbots, AI has implications that will touch nearly every aspect of our lives faster than any other technology that has ever existed throughout the course of human history. If you know us, we are not interested in theatrics or bloviating, so we do not make the previous statement lightly! DevOps, or the integration of software development and IT operations practices, has been revolutionizing the infrastructure and software industry for some time now. However, with the introduction of GPGAI applications, DevOps is currently undergoing a transformation that will fundamentally change the way software is developed and deployed forever.


As engineers and companies begin to rely more heavily on AI systems, the rise of AI generated full stack applications will create unprecedented levels of software complexity, which will necessitate a change in how the DevOps cycle incorporates new sets of technologies. Applications that implement GPGAI will be able to rebuild entire software systems from scratch, using generative adversarial networks (GANs) - two neural networks contest with each other in the form of a zero-sum game, where one agent's gain is another agent's loss. - that will produce code that not only conforms to specific design requirements, but code that is also superior to all others tested in the zero-sum-game. By applying GPGAI to DevOps, developers will be able to automate an even wider range of processes, including coding, testing, and even monitoring software performance.


One significant transformation that will most likely emerge with the introduction of GPGAI in DevOps is the new role of software developers. The AI applications themselves will likely end up writing much of the code that previously was the responsibility of humans. When that seemly inevitable transition happens, developers will find themselves responsible for overseeing performance, debugging, analysis, and optimization of these AI-generated applications. Essentially, as software development is automated, developers will be tasked with more of a strategic role within the organization. They will become more of a resource to outline value propositions associated with specific products, understand customer demands, and communicate directly with business units.


Given how GPGAI applications are able to write and test code faster than human developers, the traditional DevOps cycle has started to change. Rather than separate development, testing, and deployment phases, GPGAI will blur these distinctions, further bridging the gap between software developers and IT operations personnel. This shift will ultimately create a more continuous cycle of software development, testing, and deployment, resulting in more rapid and seamless updates to software and software-based services. This realignment of DevOps will lead to a new approach for developing software: the Intelligent Continuous Integration (ICI) pipeline. These ICI pipelines will take advantage of machine-generated code, using clustering techniques to manage numerous code iterations, and then monitoring the convergence of the network as it trains. The result is a complete software solution that is optimized through a process akin to genetic evolution - a heuristic search algorithm used to solve, search, and optimize. If you would like to learn a little more about evolutionary algorithms, check out this https://www.youtube.com/watch?v=qv6UVOQ0F44 and https://en.wikipedia.org/wiki/Evolutionary_algorithm. For those of you that had the spark of your childhood reignited by the new Mario Brothers movie, I highly recommend checking out the first link where Super Mario is used to show and explain how AI genetic evolution search algorithms work.


THE END

DevOps is poised to change in a massive way with the introduction of general purpose generative artificial intelligence applications. The integration of AI technology will result in transformations in developer responsibilities, DevOps cycles, software development practices, and innovation opportunities. Given the numerous advantages that AI brings to development, many DevOps teams are likely to embrace this coming transformation.

And while there are many upsides to the coming change, as technologists we must also seriously consider the downsides to the adoption and integration of GPGAI backed by large language models (LLMs). AI is not a tool. AI is something entirely different and its impact will not be limited to the silly and incredibly short-sited comments on social media claiming "AI will not take your job, but a person who knows AI will take your job". AI does not represent the transition from horses to cars. It just doesn't, and anyone who tells you otherwise you should be very very cautious of their intentions. Look, the reality is, the only remotely close historical reference humanity has that we can use as a comparison to the impact that GPGAI will have, is the type of massive change that occurred when the Spanish conquered South America due to their overwhelming technological superiority. If you are not aware of how crazy some of these battles were, a great example is the Battle of Teocajas, Sebastian de Benalcazar had 140 Spanish and Cañari allies: together they fought an Incan force of thousands to a draw. And, honestly, in the coming months and years for those that have AI and those that don't the gap between success and failure will be even greater than the Battle of Teocajas. The point is, AI is not a tool. AI is something entirely different, and we must tread very carefully and responsibly when integrating AI into systems.

20 views0 comments

Comments


bottom of page