The AI Revolution in Software Engineering: A Double-Edged Sword
Dr Vahid Garousi, CISC Senior Lecturer, was featured in the latest edition of Sync NI, which explored the impact of AI on software engineering.
How AI is Transforming Software Engineering: Hype, Reality, and What Comes Next
Artificial intelligence is no longer a futuristic concept — it is already reshaping the way software is designed, built, and tested. At an invited talk with technology leaders in Northern Ireland, Dr. Vahid Garousi, Senior Lecturer in Software Engineering at Queen’s University Belfast, explored how AI is changing the daily work of software engineers and how organisations can prepare for this shift responsibly.
The session was structured around four pragmatic questions that practitioners and researchers are now grappling with:
- Which software engineering tasks are already being supported by AI, and to what degree?
- How much do we trust the quality of AI-produced code, tests, and documentation?
- Is AI making engineers more productive — and could it also gradually reduce human skills?
- How can universities and industry collaborate to build an effective and safe AI-assisted engineering future?
AI’s Promise — And Its Limits
The software engineering field was formally established in 1968 at a NATO conference — born from the idea that software systems require the same rigorous, disciplined approach as civil, mechanical or electrical engineering. Today, the profession is entering its biggest transformation since that moment.
AI tools can now assist across nearly every phase of the software lifecycle — not just coding, but also requirements, design, testing, maintenance, and operations. In one of our research projects, we mapped 66 commercial AI tools for software testing alone, showing how rapidly this ecosystem is growing.
But AI is not magic. Its value depends on context, data quality, and skilled human direction. In Dr. Garousi’s words:
“AI is powerful not because it replaces engineers — but because skilled engineers learn how to guide, inspect and correct it.”
Productivity: Yes — But With a Catch
AI can accelerate certain engineering tasks, especially repetitive, pattern-driven activities such as test generation or code suggestions. Organisations that apply AI thoughtfully often see clear efficiency improvements.
However, the equation is not always positive:
- Poorly defined prompts lead to poor results
- Engineers must review and correct AI output
- “Hallucinations” — false but confidently generated statements — create real risks
- Verifying AI output can sometimes take as long as writing it manually
This creates a new paradox in engineering:
AI can save time — but only if humans stay deeply involved.
Where AI Performs Well — and Where It Doesn't
AI performs best in tasks where patterns are strong and context is limited — for example, creating boilerplate code, classifying test results, or identifying visual UI changes. Our empirical studies with industry partners showed that AI-powered testing tools can reduce maintenance effort, especially for simple interface changes.
But AI struggles where:
- Context is complex
- Creativity or strategic thinking is needed
- Safety or ethics matter
- Outputs involve ambiguity or innovation
Activities like forming a product vision, designing user experiences, or making ethical trade-offs still require a human mind. As Dr. Garousi explains, “Creating a new interface that users actually love remains a fundamentally human task — for now.”
Trust and Oversight Are Essential
Trust emerged as a central theme. While AI can generate helpful suggestions, engineers report mixed experiences with quality and reliability. Many teams now apply a “human-in-the-loop” model: AI proposes, humans evaluate.
This mirrors how society is approaching autonomous systems like self-driving cars — ranging from Level 0 (no automation) to Level 5 (full autonomy). Most AI use in software engineering today sits around “Level 2”: assistive, but requiring close review.
Regulators and policymakers are also taking note. New AI laws emphasize accountability and demand human oversight — especially for applications that affect safety, finance, healthcare or privacy.
Impact on Skills and Jobs
Media headlines often focus on whether AI will “replace all programmers.” The reality is more nuanced. Yes, some routine coding roles may shrink, and junior tasks may change. But new roles are emerging:
- AI-assisted developer
- Prompt engineer for software tasks
- AI quality auditor (“AI reviewer”)
- Human-in-the-loop systems supervisor
- AI policy and governance specialist
- AI testing and reliability engineer
Just as calculators did not eliminate mathematics, AI is unlikely to eliminate software engineering — but it is redefining the skills required.
Opportunities for Collaboration
Dr. Garousi emphasized the importance of industry–academic partnerships to navigate this transition. Areas of collaboration include:
- Building reliable AI-generated code and tests
- AI maturity assessment for software organisations
- Upskilling teams in AI-assisted engineering
- Developing responsible-AI policies for engineering teams
- Measuring productivity vs. cognitive load in AI workflows
- Human-in-the-loop quality assurance frameworks
Northern Ireland’s innovation ecosystem — including Software NI and AICC — is well positioned to shape responsible AI adoption.
Looking Ahead
AI is already contributing meaningfully in areas far beyond software engineering. It is helping model climate change, accelerate medical research, improve agriculture, and optimize energy systems. The long-term vision extends to breakthroughs such as carbon-neutral economies and faster vaccine discovery during future pandemics.
But across all domains, one principle remains constant:
AI will not replace engineers — but engineers who master AI will shape the future.
Software engineering is becoming a partnership between human expertise and intelligent tools. The challenge now is to approach this shift thoughtfully, train the next generation accordingly, and build systems that amplify — not replace — human creativity, ethics, and judgment.
Click here to read the full article here or view Dr Garousi's Access the Expert Seminar Series presentation.