According to the 2024 DORA report, which synthesises responses from nearly 39,000 developers, 76% of respondents rely on AI for tasks such as code writing, information summarisation, and code explanation.
More than three-quarters of developers now use AI for at least one daily professional responsibility.
Does this mean AI has moved beyond the hype and is now delivering real value in everyday development tasks? It certainly seems so — if not for one major caveat: 39% of respondents still report little or no trust in AI.
As always, the truth lies somewhere in between.
Meet our “heroes”
The State of Developer Ecosystem Report 2024 highlights a lineup of AI tools transforming the coding landscape, including ChatGPT, GitHub Copilot, Google Gemini, Microsoft 365 Copilot, Amazon Q, and many more.
Among them, ChatGPT reigns supreme. A staggering 69% of developers have tried it for coding and development tasks, and 49% use it actively. For comparison, GitHub Copilot comes a close second, with 40% having tested it and 26% using it regularly.
Interestingly, Vention’s developers mirror this global trend: ChatGPT leads the pack, closely followed by GitHub Copilot. My personal favourite right now is Cursor. I’ve been using it on my pet projects, and it’s been genuinely helpful. It helps me move faster. And, importantly, it gets things about 90% right. That makes a huge difference when you’re experimenting, iterating, or just building for the fun of it.
The bright side of AI copilots
AI assistants shine in four key areas:
Code generation
Developers can use open-source language models to speed up coding and simplify workflows. These tools generate usable code from simple instructions and automate repetitive tasks like boilerplate coding. The benefits go beyond time savings — AI ensures consistency and alignment with a project’s coding style.
Code reviews and debugging
AI assistants scan code to identify redundancies, syntax errors, logic flaws, and runtime issues. But they don’t stop there — they also assist with debugging. For example, if a critical function is missing a null check, an AI tool can pinpoint the exact line of code that needs fixing.
Refactoring and optimisation
AI can suggest structural improvements and best practices to enhance code quality. It detects redundant code and recommends reusable functions or modules to reduce duplication. Additionally, AI analyses project dependencies, flagging outdated or unnecessary libraries to keep the codebase clean and efficient.
Documentation
By analysing functions, classes, and APIs, AI can generate documentation that explains what the code does, its parameters, and expected outputs. The best part? It can automatically update documentation whenever the code changes, ensuring everything stays in sync.
Vibe coding
Vibe coding is an AI-driven programming approach that enables users, including those without extensive coding skills, to create software by simply describing their desired outcomes using natural language. It accelerates development, can boost productivity, and lowers the barrier to entry by allowing users to focus on high-level concepts rather than routine coding tasks.
However, the dark side of these tools can be more extreme with this approach.
The dark side of AI copilots
Ironically, the disadvantages are closely related to the advantages.
The risk of lower-quality code
Yes, despite all the amazing advantages, AI assistants aren’t flawless. They still struggle with reasoning, and the quality of their suggestions hinges on the quality of their training data. This means AI can — and sometimes does — generate inefficient, outdated, or insecure code.
The good news is that AI models are constantly evolving. OpenAI, for example, has promised that its next iteration will offer improved reasoning capabilities.
Still, blind trust is risky. Expert review and critical thinking remain non-negotiable. Junior developers are particularly vulnerable — without experience, they may not always recognise when AI gets it wrong.
Complexities during debugging
AI-generated code can be complex or contain unfamiliar constructs, requiring developers to spend extra time understanding it before debugging.
There’s also the risk of over-reliance. If developers lean too much on AI, they may miss certain bugs, shifting the burden to QA teams, who then have to handle code that could be worse than manually written alternatives.
Remember: while each AI assistant has its strengths, none can replace human expertise. These tools should enhance workflows, not replace critical processes such as code review and decision-making.
Hesitancy from engineers
The promise of AI-driven efficiency is often shadowed by concerns:
- Will AI replace my job?
- Are my skills deteriorating because I rely on AI?
- Can I trust AI with sensitive information and NDAs?
The State of Developer Ecosystem Report sheds light on this scepticism. Only 29% of respondents said that cloud-based AI tools are allowed for all projects, while 11% noted that they’re prohibited for all projects.
And I’m not alone in my thoughts. When we asked Vention’s engineers about the role of AI in software development, here’s what they had to say:
- “AI still needs a supervisor.”
- “Even if you like AI’s output, the system won’t explain how it arrived at that decision.”
- “AI could assist engineers in the same way that calculators help mathematicians perform complex calculations.”
AI is the future — no doubt about it
The real challenge lies in using it wisely.
We’re now in the third year of the AI boom, with AI adoption rates skyrocketing and AI-driven innovation reshaping industries. Yet, its dual nature remains evident — it can drive efficiency just as easily as it can create new challenges.
At Vention, we’re passionate about AI and continuously explore its full potential. We don’t just use AI; we put it to the test across the entire software development lifecycle, from planning and coding to deployment and ongoing support, all while ensuring zero risk to our clients’ projects.
And we’re committed to using AI responsibly. That’s why we’ve introduced the Vention AI Manifesto, a framework that lays out our guiding principles for ethical, human-centred AI.
I firmly believe this is the key to success: ensuring that every team member — not just developers — becomes proficient in AI, with prompt engineering emerging as a critical skill.
