Artificial intelligence capable of producing software code has not yet rendered engineers at Anthropic obsolete. During a thoughtful and forward-looking discussion held on Wednesday with Salesforce’s CEO, Marc Benioff, at the widely attended annual Dreamforce conference, Anthropic’s co-founder and CEO, Dario Amodei, shed light on how his organization is harnessing AI for software development. He explained that while Claude, Anthropic’s AI model, now generates approximately ninety percent of the code for most development teams within the company, human engineers remain absolutely indispensable to the process. Amodei noted that he had once predicted that within six months, a vast majority of coding—close to ninety percent—would be executed by AI systems. Although some observers initially regarded this forecast as overly optimistic or even unrealistic, Amodei affirmed that within Anthropic itself, as well as in several companies partnering with them, this transformative prediction has already become a practical reality.
Prompted by this revelation, Benioff inquired how soon that percentage might increase even further and whether such progress implied a reduced need for software engineers. In response, Amodei emphasized the importance of not misconstruing Claude’s remarkable capabilities. He cautioned that the AI system’s ability to effortlessly design new features or resolve complex, persistent bugs does not equate to the displacement of human professionals. Instead, these advances alter the nature of the work engineers perform. As Amodei explained, when Claude writes ninety percent of the code, that typically signifies not redundancy, but rather a reconfiguration of human roles. Companies often still require the same number of engineers—or possibly even more—because these individuals now act as strategic multipliers of productivity. The AI handles most of the routine and mechanical aspects of programming, while human experts concentrate their efforts on the intricate ten percent of tasks that require creativity, judgment, and supervision of multiple AI systems. According to Amodei, this synergistic arrangement enables engineering teams to expand their impact exponentially, resulting in productivity gains up to an order of magnitude greater than before.
Amodei summarized this shift as a process of rebalancing rather than replacement, underscoring that AI integration is reshaping responsibilities instead of eliminating them. His remarks resonate with broader trends across the technology sector. Anthropic’s experience is far from unique: organizations throughout the industry are experimenting with similar AI-driven workflows. For example, in March, Garry Tan, president and CEO of the renowned startup accelerator Y Combinator, remarked on social media that roughly one-quarter of the founders in Y Combinator’s 2025 winter batch were already producing as much as ninety-five percent of their code with the assistance of AI tools. This trend highlights an accelerating transformation in how software is conceived and developed globally.
However, this rapid adoption of automated coding technologies is not without consequences for the labor market. A recent study conducted by Stanford University found that the widespread use of AI programming assistants is already reshaping entry-level opportunities within the software engineering profession. The research revealed a notable twenty percent decline in employment among developers aged twenty-two to twenty-five between late 2022—the period coinciding with the release of ChatGPT—and July 2025. This contraction raises concerns about potential disruptions to the traditional pipeline of emerging technical talent, as fewer young professionals may choose to pursue careers in fields where machines appear to dominate foundational tasks. In contrast, workers with greater experience have remained relatively insulated from the immediate effects of this technological upheaval, suggesting that expertise, strategic reasoning, and high-level oversight continue to hold enduring value even in an era increasingly defined by automated code generation. Collectively, these developments depict a nuanced future—one in which human ingenuity and artificial intelligence coexist in a delicate balance, each reinforcing and amplifying the other’s strengths rather than competing for dominance.
Sourse: https://www.businessinsider.com/most-anthropic-teams-coding-with-claude-ai-not-replacing-humans-2025-10
