AI Writes 41% of Code. Only 29% of Devs Trust It. Review It Like a Senior Engineer
A new report shows developers now trust AI-written code less than before, with studies finding it slower and more vulnerable to flaws.
- Trust in AI-generated code has declined from 40% to 29% in one year, per Stack Overflow data.
- AI code is 19% slower and 2.74x more likely to contain vulnerabilities, according to Veracode and METR.
- The bottleneck in software development has shifted from code generation to verification and quality assurance.
- Experts recommend a 10-point review checklist to mitigate risks in AI-assisted coding.
Recent data from Stack Overflow, METR, and Veracode reveals a growing skepticism toward AI-generated code among developers. Trust in AI-written code has dropped from 40% to 29% in just a year, while performance benchmarks show AI code is 19% slower than human-written alternatives. Worse, Veracode’s analysis found AI-generated code is 2.74 times more likely to contain vulnerabilities, highlighting a critical shift from code generation to verification as the primary bottleneck in software development.
The trend underscores a broader challenge: while AI tools like GitHub Copilot and others have accelerated coding, they’ve also introduced new risks. Developers are now advised to adopt rigorous review processes, with experts suggesting a 10-point checklist to catch flaws early. This shift reflects a maturing phase in AI-assisted development, where quality control and security take precedence over raw productivity gains.
Highlights the need for stricter code review processes when using AI tools.
Underscores the importance of auditing AI-generated code for security and performance before deployment.
Serves as a cautionary example of balancing AI tools with critical evaluation in software engineering.
- METR
- A research group focused on measuring AI performance and reliability in software development.
- Veracode
- A security firm specializing in application vulnerability detection and remediation.
AI ToolsEurope's free satellite service just made it easier to track wildfires
AI ToolsDurable approval is not the same as valid approval
AI ToolsAnthropic loosens Fable 5's biology restrictions but keeps the guardrails on for virology and toxicology
How Cohere Health digitizes clinical policies using Amazon Bedrock AgentCore - Amazon Web Services (AWS)
How TReNDS automates root-cause analysis with Amazon Bedrock - Amazon Web Services (AWS)
NeonatAIlogy: The GIRISH (Goal, Input, Role, Iterative Refinement, Safety Verification, and Human Accountability) Framework for Structured, Human-Accountable Artificial Intelligence in Neonatal Intensive Care - Cureus
Researchers propose the GIRISH framework for human-accountable AI in neonatal intensive care. The framework aims to ensure safe and effective AI integration in healthcare.
Oneida County summer youth program introduces students to AI - Rome Sentinel
Oneida County's summer youth program is introducing students to artificial intelligence. The program aims to educate students about AI and its applications.
OpenAI says it slowed Astra model development over security concerns
OpenAI has temporarily halted parts of its Astra AI model development because of security vulnerabilities. The move reflects growing scrutiny of AI safety in advanced models.
In the News: John Abraham Discusses AI Safety Concerns - Newsroom | University of St. Thomas
John Abraham, a prominent AI safety advocate, shares his thoughts on the pressing concerns surrounding AI development in an interview with the University of St. Thomas.
Who is liable when Artificial Intelligence goes rogue? - FOX 29 Philadelphia
The issue of liability when artificial intelligence systems fail or cause harm is becoming increasingly important. Experts are debating who should be held responsible in such cases.
BusinessAfter Rippling blew millions on AI in months, it built an employee ROI tool
Rippling introduces AI Spend Console to monitor AI tool spending by employees and teams, addressing cost inefficiencies after rapid AI adoption.