We are not going to agree on AI (idiallo.com)

The ongoing debate about the effectiveness of AI in development circles has been highlighted by contrasting opinions among developers, where one successfully produces 30,000 lines of code monthly using AI, while another remains skeptical, branding it ineffective. This divergence reflects a broader uncertainty in the AI/ML community about the true capabilities and reliability of AI technologies. Despite reports of companies like Medvi generating significant revenue—projected at $1.8 billion this year—there are concerns over the integrity and impact of AI-generated code, exemplified by Microsoft’s claims that 30% of its codebase is AI-generated, juxtaposed with GitHub logging numerous incidents related to coding errors. The article emphasizes that consensus on AI's utility may be elusive, as opinions vary widely. It underscores the notion that AI should be viewed as a tool—capable, yet imperfect—and its true value lies in how effectively it enhances individual workflows. The author recounts a personal experience in a Home Depot that serves as an allegory, suggesting that whether one opts for a simple or advanced tool depends on specific needs and goals. As AI continues to evolve, the crux of the matter remains: its performance and reliability depend on the user's understanding and application of these technologies, not on the hype surrounding them.