GenAI is being hailed as a revolutionary coding tool. Yes, it creates huge opportunities for development teams, but we must remember that AI is a junior developer, not an engineer.
The idea that AI will take over app development overlooks a central aspect of a developer’s job. There’s a reason we call them developers or engineers and not typists. Writing commands has never been the hard part. The challenge lies in ensuring that the code solves the problem at hand within the constraints and domain of the product. The current generation of GenAI does not achieve this.
The role of the developer is not dead: it is evolving. As AI assists in code generation, the creativity, strategic thinking and contextual understanding of human developers will be even more crucial in shaping successful software solutions.
Vice President of Engineering at Prismatic.
The limitations of GenAI in software development
Developers have long used code templates, generators, and autocomplete functions to speed up programming. GenAI can take these tools a step further by writing entire functions or blocks of code from natural language prompts. However, AI does not fully understand logic and lacks context about the business problems and purpose of the software, resulting in mediocre code.
For example, GenAI can create code that calculates total sales revenue. However, the result may not take into account organization-specific variables, such as including returns and refunds in the equation and formatting the results to meet reporting requirements. The code works technically, but it doesn’t actually solve the problem.
Additionally, GenAI tools often generate incorrect code. The training data for large language models (LLM) contains high-quality and low-quality data, and the algorithm cannot decipher the difference. Research from Bilkent University measured performance in terms of code quality metrics and found that ChatGPT only wrote correct code 65% of the time, while GitHub Copilot and Amazon CodeWhisperer performed even worse.
AI-generated code can also introduce vulnerabilities and compromise data security by not following security protocols. This risk is made more dangerous by many developers’ mistaken trust in algorithms.
A Stanford University study found that developers who used AI to write code were more likely to believe it was secure when, in fact, they were less so than teams that didn’t use an AI tool. These results suggest that programmers may become less attentive when reviewing their work as a result of relying on AI. More than 90% of security leaders are concerned about the use of AI in encryption, but less than half have policies in place to ensure its secure use.
In light of these challenges, experienced human developers will always be needed in application development.
What is the developer of the future like?
Gartner projects that 90% of enterprise software engineers will use AI code assistants by 2028, moving developers into strategic advisory roles. However, the core responsibilities of developers—maintaining code quality, strategically adapting systems to changing environments, and meeting specific project demands—will remain essential.
Developers and engineers will increasingly act as architects, specifying high-level requirements and constraints while AI completes the detailed coding. This means developers should focus less on writing low-value, low-context code and more on understanding business requirements, system architecture, edge cases, and performance testing.
The cooperative relationship between AI and humans could resemble pair programming. AI will play the role of a less experienced partner performing basic tasks, allowing developers to spend more time guiding and suggesting improvements to the code.
AI integration could push development teams to shift further left on traditional code review practices such as linting, testing, and compliance checks. Because GenAI can produce functional but contextually inaccurate or insecure code, incorporating checks early in development allows teams to proactively detect issues. This approach improves code quality, reduces the risk of errors, and maintains consistency.
While GenAI can offer many benefits, it presents a conundrum for the professional sector. Since AI works as a junior developer, companies may need to hire fewer entry-level developers. This situation limits opportunities for human employees to improve their skills, resulting in fewer people equipped to monitor code quality. This scenario remains an unsolvable problem, a problem that needs an answer soon.
Developer Fundamentals Will Last
The value of a developer lies in understanding the broader purpose and structure of the code, not just the act of writing it. Basically, GenAI will not alter the skills needed for this job, although developers can spend less time with their hands on the keyboard. Critical thinking and adaptability will be even more essential for success. Since AI handles most of the tedious tasks, developers must master the skills to instruct and correct AI to achieve the desired result.
We have introduced the best web development tool.
This article was produced as part of TechRadarPro’s Expert Insights channel, where we feature the best and brightest minds in today’s tech industry. The views expressed here are those of the author and are not necessarily those of TechRadarPro or Future plc. If you are interested in contributing, find out more here:
