Andrej Karpathy, a prominent figure in artificial intelligence and a former director of AI at Tesla, has raised significant questions regarding the evolving landscape of software engineering. He suggests that the integration of AI into coding practices is not just a trend but a fundamental shift in how software is developed. Karpathy has coined the term ‘vibe coding’ to describe this new approach, where coding is increasingly driven by AI tools rather than traditional manual methods.

According to Karpathy, his own coding style has transformed dramatically, with 80% of his coding now performed by AI agents, leaving only 20% for manual coding. He notes, “I really am mostly programming in English now, a bit sheepishly telling the LLM what code to write… in words.” This shift highlights a growing reliance on AI tools, which have become more coherent and capable over time.

Karpathy points to a pivotal moment in December 2025, when AI coding agents reportedly crossed a threshold of coherence, marking a phase shift in software engineering. This change has been echoed by other developers in the field. Boris Cherny, a developer, stated, “Pretty much 100% of our code is written by Claude Code,” underscoring the extent to which AI is now integrated into the coding process.

The impact of AI-assisted coding is evident in recent surveys. A Stack Overflow 2025 Developer Survey revealed that 84% of respondents either use or plan to use AI-assisted programming. However, this enthusiasm is tempered by skepticism; 46% of developers expressed doubts about the accuracy of AI tools, while only 33% felt confident in the results produced by these systems.

One notable project in this evolving landscape is NanoClaw, developed by Gavriel Cohen, which has garnered over 18,000 stars on GitHub. Despite its popularity, Cohen’s legitimate site is being overshadowed by a fraudulent site in search rankings, highlighting the challenges that developers face in navigating the AI-driven coding environment.

Karpathy’s reflections on AI-assisted coding indicate a significant change in software engineering practices. He acknowledges a personal struggle with this shift, stating, “I’ve already noticed that I am slowly starting to atrophy my ability to write code manually.” This sentiment resonates with many developers who are adapting to a new reality where AI tools play an increasingly dominant role.

As the software engineering community continues to grapple with these changes, the future of coding remains uncertain. While the benefits of AI-assisted programming are clear, the ongoing concerns about accuracy and reliability suggest that the transition will require careful consideration and adaptation from developers. Details remain unconfirmed regarding how these dynamics will evolve in the coming years, but the influence of figures like Karpathy will undoubtedly shape the discourse around AI in software development.

The post Andrej Karpathy and AI-assisted coding appeared first on bangalinews.

Andrej Karpathy, a prominent figure in the AI landscape, asserts that the integration of AI into software engineering is not just a trend but a transformative phase. He emphasizes that his own coding practices have evolved to a point where approximately 80% of his coding is now done through AI agents, with only 20% remaining as manual coding. This shift underscores a significant change in how developers approach programming.

AI’s Impact on Coding Practices

Karpathy, who previously served as the Director of AI at Tesla and is a founding member of OpenAI, coined the term “vibe coding” to describe the new paradigm of AI-assisted programming. He remarked, “I really am mostly programming in English now, a bit sheepishly telling the LLM what code to write… in words.” This highlights the growing reliance on language models to generate code, reflecting a broader trend in the software development community.

Developer Sentiment Towards AI Tools

The adoption of AI tools in programming is evident in a recent Stack Overflow Developer Survey, where 84% of respondents indicated they use or plan to use AI-assisted programming. However, the survey also revealed a level of skepticism among developers, with 46% expressing doubts about the accuracy of AI tools, while only 33% felt confident in the results they produce. This mixed sentiment illustrates the ongoing debate regarding the reliability of AI in coding.

The Emergence of New Projects

In early February 2026, Gavriel Cohen launched the NanoClaw project, which has quickly gained traction, amassing over 18,000 stars on GitHub. Karpathy praised the architectural design of NanoClaw, recognizing its potential impact on the software engineering landscape. However, the project faced challenges when a fraudulent site, nanoclaw.net, was registered shortly after its launch, disseminating inaccurate information about the project. Cohen has since filed takedown notices with Google to address this issue.

Shifts in Developer Skills

Karpathy’s observations extend beyond the tools themselves; he noted, “I’ve already noticed that I am slowly starting to atrophy my ability to write code manually.” This sentiment reflects a broader concern among developers regarding the long-term implications of relying heavily on AI for coding tasks. As AI tools become more prevalent, the skills of seasoned developers may evolve, raising questions about the future of manual coding expertise.

AI’s Role in Software Engineering’s Future

As AI coding agents continue to advance, Karpathy believes they crossed a significant threshold of coherence around December 2025, marking a phase shift in software engineering. This evolution suggests that the role of AI in coding will only expand, potentially reshaping the entire industry. Boris Cherny, another developer, noted that “pretty much 100% of our code is written by Claude Code,” further emphasizing the reliance on AI tools.

Looking Ahead

While the integration of AI into software engineering presents exciting opportunities, it also raises important questions about the future of the profession. Developers must navigate the balance between leveraging AI tools and maintaining their coding skills. As the landscape continues to evolve, the implications of these changes remain to be fully understood, and details remain unconfirmed regarding the long-term effects on the developer community.

The post Andrej karpathy: The Future of AI in Software Engineering appeared first on bangalinews.