Skip to content
Accueil » AI for Coding? 46% of Developers Doubt Its Reliability, But Adoption is Growing

AI for Coding? 46% of Developers Doubt Its Reliability, But Adoption is Growing

dev_ia_coding

Artificial intelligence is increasingly establishing itself as an indispensable tool in the world of software development.

According to Stack Overflow’s 15th annual report, published on July 29, 2025, and based on responses from over 49,000 developers across 177 countries, AI is gaining ground in coding and learning practices.

It’s undeniable that with the advent of artificial intelligence, particularly thanks to our tools like the coding AI on Yiaho, it’s becoming increasingly simple and fast to write, improve, and correct lines of code. However, despite this growing adoption, often called “vibe coding,” a certain distrust persists regarding the reliability of AI-generated results.

This article, written by the Yiaho team, explores the study’s key figures, emerging trends, and implications for the future of development.

AI for Coding: Rising Adoption, But Still a Minority

According to the report:

  • 31% of developers currently use AI tools for coding,
  • 17% plan to start using them soon.
  • 69% of developers have not yet integrated AI into their workflow, a figure that reflects cautious adoption!

Compared to 2024, when only 24% used AI (76% non-users), the evolution is notable, but AI is far from omnipresent.

This progression is explained by the democratization of tools like GitHub Copilot, Codeium, and AI assistants integrated into development environments. These tools promise to accelerate code writing, automate repetitive tasks, and suggest solutions. However, their integration is still hampered by doubts about their effectiveness and accuracy.

Learning to Code with AI: A Rapidly Growing Trend

AI doesn’t just assist experienced developers; it also plays a growing role in learning to code.

44% of respondents report using AI tools to learn to code, a significant increase from 37% in 2024. Platforms like Yiaho, ChatGPT, Gemini, Grok, and specialized assistants offer interactive explanations, code examples, and real-time corrections, making learning more accessible, especially for beginners.

This rise reflects a shift in learning methods: AI is becoming a virtual tutor, capable of adapting to individual needs. However, this growing dependence raises questions about the quality of knowledge acquired and learners’ ability to develop a deep understanding without relying solely on automated suggestions.

Also read on this topic: AI for Free Coding: Here’s How!

Trust in AI: A Major Obstacle!

While AI adoption is progressing, trust remains a challenge.

Among developers using AI for coding (84% of respondents in this subgroup), 46% express doubts about the accuracy of the generated results. This figure marks an increase compared to 2024, when 31% of AI users (out of 76% of respondents) were distrustful.

This growing distrust could be explained by frequent errors in code suggestions, unoptimized solutions, or biases in AI models, which require constant human verification.

Developers often point to issues such as bugs in generated code, outdated solutions, or recommendations unsuited to the specific context of their projects. These limitations remind us that, despite its advancements, AI remains an assistance tool and not an infallible solution.

AI, a Threat to Developer Jobs? Not Yet!

Despite the rapid progress of AI, developers remain confident in the future of their profession. 64% of them do not perceive AI as a threat to their jobs, although this figure is slightly down from 68% in 2024.

This confidence is based on the belief that AI, while useful for automating certain tasks, cannot replace creative reasoning, complex problem-solving, and understanding business needs, which remain essential human skills.

The slight decline in this percentage indicates a growing awareness of AI’s capabilities. Will AI replace developers? They might start considering a future where AI handles more complex tasks, forcing them to specialize further or focus on strategic aspects of development.

AI, a Promising but Imperfect Ally for Coding

The Stack Overflow study highlights a nuanced reality: AI is gradually integrating into software development, both for coding and learning, but it does not yet inspire complete trust.

With 46% of developers doubting its reliability and a majority remaining outside these tools, AI still has to prove itself. However, the progress observed between 2024 and 2025 shows growing adoption, driven by increasingly powerful and accessible tools. For developers, the future may lie in intelligent collaboration with AI: a partnership where humans maintain control, leveraging automation capabilities while compensating for its shortcomings.

As AI continues to evolve, the coming years will be decisive in determining whether it becomes an indispensable tool… or remains a simple complement in programmers’ toolboxes.

Source: Stack Overflow

Leave a Reply

Your email address will not be published. Required fields are marked *

Glen

Glen