How Vibe Coding, AI Copilots are Redefining Software Development

Juan M. Carrillo de Gea, Associate Professor, Faculty of Computer Science of the University of Murcia, Spain in an interaction with the higher education review shared his views on what “vibe coding” is and how AI copilots changing the way software is built today, how generative AI tools and low-friction development environments accelerate product development and innovation, and more.

Juan M. Carrillo de Gea is an Associate Professor at the Faculty of Computer Science of the University of Murcia, Spain. He serves as the Academic Coordinator of the Master's Degree in Big Data Analysis Technologies. His research expertise focuses on software engineering, human factors, sustainability, and the evolution of distributed development environments. With a research career spanning over 15 years, he has co-authored more than 80 scientific publications in JCR-indexed journals and CORE-ranked conferences.

What is “vibe coding,” and how are AI copilots changing the way software is built today?

“Vibe coding” represents a paradigm shift where the emphasis moves from writing formal syntax to co-creating software through natural language interaction and iterative refinement. We identify this as an emerging paradigm where conversational AI interfaces become collaborators or even autonomous agents rather than just “tools”. Today, AI copilots are changing software construction by intervening across the entire lifecycle, from requirements elicitation to testing and maintenance, effectively acting as a new, digital “team member” that can partially substitute or augment human knowledge.

How do generative AI tools and low-friction development environments accelerate product development and innovation?

These tools accelerate innovation by drastically reducing the time spent on “mechanical” or “rutinary” tasks. AI-mediated workflows can automate many tasks, from writing documentation to detecting inconsistencies in real-time. In addition, these systems are increasingly effective in coding new features without our direct intervention. This “low friction” aspect means that ideas can be transformed into prototypes very quickly, allowing teams to focus on high-level design rather than boilerplate code.

What skills will future software engineers need as AI copilots handle more coding tasks?

As AI handles the “how”, engineers must focus on the “why” and the “what”. The skills of the future will be more focused on human-centered design and organizational ergonomics. Future engineers will need deep domain expertise, the ability to critically review AI outputs for reliability, and a strong foundation in traceability to ensure compliance in regulated domains like privacy. Moreover, they will need to manage the “human dimension”, understanding how AI affects team cohesion, personal motivation and workforce well-being.

Also Read: Resurgence of Spain as a Premier International Higher Education Destination

Does vibe coding lower the barrier to entry for non-traditional developers and founders?

Absolutely. By allowing natural language to drive the development process, vibe coding effectively reduces the technical “digital divide”. Intuitive interfaces provide a gateway for individuals without traditional computer science backgrounds to build structured domain descriptions and software artifacts. However, while it democratizes creation, “expert oversight” remains indispensable. Human intelligence is still required to maintain the long-term sustainability and governance of the software ecosystem to avoid unmanageable technical debt.

How are AI-powered development tools reshaping code quality, security, and technical debt?

The impact is dual-sided. On one hand, AI can enhance requirement completeness and support automated refactoring for better energy efficiency, what we term “green refactoring”. On the other hand, we face challenges like hallucinations and the potential for increased technical debt if AI-generated code is not rigorously validated. It is necessary to align AI advances with emerging quality standards (like ISO/IEC 42042 or AI quality standards) to ensure that automation does not lead to a “cognitive debt” or unmanageable codebases.

What ethical, reliability, and accountability challenges arise when AI generates production code?

We have identified several critical “human factors” that are often overlooked. Excessive reliance on AI can lead to “cognitive offloading”, a loss of trust, and even diminished creativity among developers. There is also a significant risk regarding the “sense of ownership”; if an engineer feels the AI did the work, their commitment to the final product’s reliability and accountability might decrease. Ethically, we must also consider environmental sustainability: the energy footprint of training and running large AI models is substantial (with reports showing sharp increases in emissions and energy use), and a responsible engineer must now account for the “Green IN-AI” dimension.

Will AI copilots replace junior developers - or redefine engineering roles altogether?

Our hypothesis is not one of replacement, but of profound redefinition. While AI can handle many tasks traditionally assigned to junior developers, it creates a new necessity for “AI-augmented” roles. We are moving toward a model where every engineer acts as a “governor” of intelligent systems. The roles are shifting toward strategic integration, where the value lies in managing the complex interplay of coordination, efficiency, well-being, and sustainability. Junior roles will likely evolve into “learning-fast” positions where mastering human-AI collaboration becomes the primary gateway to the profession.