AI Now Ubiquitous in Software Development, but 90% of Firms Report Downsides
New research from Techreviewer.co shows nearly all software companies see productivity gains from AI, yet the vast majority struggle with security and reliability.
Artificial intelligence has become a standard fixture in professional software engineering, though its integration is creating a volatile trade-off between speed and stability. A new study from Techreviewer.co reveals that 89% of software development companies have now integrated AI into their coding processes to write or assist with code.
The findings highlight a stark contradiction in the current state of the industry. While 97% of these companies report tangible productivity gains from AI usage, this efficiency comes with a significant cost to reliability. According to the research, 90% of companies report encountering at least one significant downside, specifically citing the emergence of security vulnerabilities and "hallucinated" code—instances where the AI generates syntactically correct but logically flawed or non-existent functions.
The Shift to Automated Development
This widespread adoption comes during a broader industry transition toward automated coding agents and integrated development environments (IDEs) that embed large language models (LLMs) directly into the workflow. Techreviewer.co, a B2B research and analytics platform founded in 2019, tracks these trends by evaluating software engineering firms on technical expertise and project success rates. The data suggests that AI is no longer an experimental tool for early adopters but is now a core component of the modern development stack.
The Cost of Efficiency
The tension between output and accuracy suggests that AI has not yet reached a state of safe autonomy. The high prevalence of reported downsides indicates that while AI can accelerate the drafting of code, it cannot yet be trusted to ensure the integrity of a production environment. The prevalence of security flaws and logic errors means that human oversight is not merely a preference but a critical requirement to prevent catastrophic failures in live software.
The Path Forward
As companies continue to lean on AI to meet tighter deadlines, the industry must now pivot from focusing on raw productivity to developing more robust verification frameworks. The primary challenge remaining is the mitigation of hallucinations and the closing of security gaps introduced by automated tools. Until these reliability issues are solved, the role of the human developer will shift from primary author to a critical auditor of AI-generated output.