Now’s the Time: Computer Science Must Evolve to Emphasize Software and Systems Engineering with Artificial Intelligence (AI)

Chandra N. Sekharan, George K. Thiruvathukal · 2026

Read Now’s the Time: Computer Science Must Evolve to Emphasize Software and Systems Engineering with Artificial Intelligence (AI) by Chandra N. Sekharan, George K. Thiruvathukal (2026) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2604-27230.txt.

Introduction

Sekharan and Thiruvathukal argue that computer science teaching must change now to center software and AI systems engineering. Old curricula treat coding, data structures, and algorithms as ends in themselves. Instead those topics should serve as building blocks for systems-centered training. Graduates should design, orchestrate, verify, and own complex AI-enabled systems under real-world limits, rather than compete with AI on routine coding tasks.

Catalog record

CREATORChandra N. Sekharan, George K. Thiruvathukal
YEAR2026
COLLECTIONPAPERS
FORMATTXT
FILE SIZE17119
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2604.27230

Browse the full library · About the Tower of Babel

Full text

The opening of the work, from the library's hosted text:

Now’s the Time: Computer Science Must Evolve to Emphasize Software and Systems Engineering with Artificial Intelligence (AI) Chandra N. Sekharan, George K. Thiruvathukal Abstract Computer science (CS) education needs to evolve to support software and artificial intelligence (AI) systems engineering, and it needs to happen now—precisely because the core intellectual contributions of CS have never been more important. We argue that traditional curricula, built around programming, data structures, and algorithms as ends in themselves, must be reframed so that these topics become foundational building blocks within a systems- and engineering-centered education. Graduates should be prepared not to compete with AI on routine coding tasks, but to design, orchestrate, verify, and own complex AI-enabled systems operating under real-world constraints. More importantly, computer science education should be geared towards preparing students for future disruptions. The broad history of computing is marked by one disruptive technology after another, requiring us to rise to the moment instead of merely acquiescing to it. K eywords computer science education \cdot artificial intelligence \cdot software engineering \cdot systems engineering \cdot curriculum reform Imagine a civil engineering graduate who has spent four years learning to lay bricks, mix concrete, and swing a hammer with precision—but has never designed a bridge, calculated load-bearing stresses, or managed a construction project where real lives and real money are at stake. Mechanical engineers do not personally weld every joint in a factory line, and electrical engineers do not hand-solder every production board. They master the fundamentals so they understand materials and processes intimately, then spend their careers architecting, integrating, verifying, optimizing trade-offs, ensuring reliability, and owning delivery of complex, real-world systems. The manual crafts are left to skilled tradespeople, or increasingly to robots, working to their specifications. Computer science education needs the same evolution. For decades, CS degrees have drilled first-year students on Python or Java syntax, then marched them through data structures, computer organization, and algorithms as sacred rites of passage. Students grind out “LeetCode”-style problems, implement canonical data structures, and then move on to a mix of theory, operating systems, databases, networking, and advanced mathematics. We regard this body of material as essential: students still need to recognize a data structure when they see it, understand its performance and memory trade-offs, and know why a particular algorithm or representation is appropriate in a given context. What must change is the way these topics are framed—not as ends in themselves, but as reusable design primitives and building blocks in larger systems that students will rarely, if ever, implement entirely from scratch again. The traditional and current curriculum took inspiration from computational thinking (CT) [ 1 ] , which, as Jeannette Wing argued, anchored the field with a powerful vocabulary for abstraction, decomposition, and algorithmic design. Properly interpreted, CT actually includes systems-oriented concerns—prefetching and caching, redundancy, damage containment, error correction, race conditions, deadlock, separation of concerns, and trade-offs between time, space, processing, and storage are all there. The real gap is that many curricula seem to have adopted a narrow interpretation of CT, stopping at algorithms and abstraction while largely downplaying the systems, uncertainty, and engineering dimensions that were present all along [ 5 ] . The result is familiar to anyone hiring: graduates who can write elegant code in isolation or in small teams, but do not always think through the more holistic considerations of shipping, scaling, securing, and maintaining systems at scale. While CS programs have changed only incrementally, enterprises and alternative providers have created parallel pathways for tech-savvy students to learn in more applied, systems-oriented settings that allow them to go straight into work. The challenge is not that CS theory, algorithms, and data structures are obsolete; it is that they are not yet being consistently taught as the conceptual “hardware” or “building blocks” of modern software and AI systems. AI Has Already Moved the Goalposts Generative AI has now thrown this slow drift into overdrive. Tools like Cursor, Claude, Gemini, Copilot-style assistants, and emerging agentic platforms already generate, debug, and refactor production-grade code at speeds humans cannot match for many classes of problems. We acknowledge that these tools are far from perfect, but their rate of improvement is unmistakable, and they will continue to improve at a rapid pace. We joke with our students that if you sleep for too long, a new or improved AI model might be there when you wake up. The irony is that the very systems thinking we are arguing for is what made modern AI possible in the first place. Large-scale AI is not the product of a single clever algorithm; it is the outcome of systems thinking at massive scale—breakthroughs in distributed systems, data pipelines, and hardware integration, layered with careful engineering in pretraining, fine-tuning, and reinforcement learning from human feedback (RLHF) to transform a stochastic next-token predictor into one of the most useful artifacts in contemporary tooling. This massive scale itself is beyond the capabilities of most institutions and is now the province of just a handful of companies that were pioneers in cloud computing. The industry is already reorganizing around this reality. Gartner, for example, projects that within a few years the overwhelming majority of enterprise software engineers will be using AI code assistants on a daily basis [ 2 ] , with roles shifting from “implementation” to “system design and orchestration”. McKinsey & Company’s 2025 work on AI in software development finds that the highest-performing organizations see sizeable gains in productivity, time to market, customer experience, and software quality—on the order of tens of percentage points—by embedding AI across the entire product-development lifecycle and redefining roles and practices accordingly [ 3 ] . Early-career roles are already feeling the pressure: recent work from the Stanford Digital Economy Lab reports that since the widespread adoption of generative AI, early-career workers (ages 22–25) in the most AI-exposed occupations, including software developers, have experienced a considerable relative decline in employment, with adjustments occurring primarily through reduced hiring rather than wage cuts [ 4 ] . Routine programming is turning into a commodity, not because software is going away, but because the nature of software work is changing. We have been here before, at smaller scales. Mainframe computers and personal computers (PCs) were supposed to wipe out office work; instead, they reshaped it. The World Wide Web was going to destroy “brick and mortar”; instead, it rewired commerce; “brick and mortar” remains and likely will do so in an AI-enabled world. Wikipedia was feared as a flattening force that would replace experts; instead, it became part of the foundational infrastructure—and training data—for today’s AI, which basically relies on a snapshot of the Internet as a crucial aspect of its training. Each time, computing reinvented itself as a distributed, networked, systems discipline. The internet’s evolution into the web and then into cloud computing is, at heart, a story of systems and scale. This latest AI wave is consistent with that pattern but more intense, and it demands that we raise the level of abstraction and awareness in what we teach beyond the basic coding tasks that were already influenced by information available from sites like Stack Overflow and Chegg long before large language models (LLMs) arrived. Suffice it to say, not only has

Read the complete text