People training new AI models admit they just get chatbots to do it
The next generation of AI models are meant to be trained by people paid to have conversations with them, but several of these workers have admitted to New Scientist that they simply get chatbots to do
The next generation of AI models are meant to be trained by people paid to have conversations with them, but several of these workers have admitted to
Read Full Story at New Scientist →Why This Matters
The revelation that workers training cutting-edge AI models are outsourcing parts of their labor to chatbots exposes a critical flaw in the industry’s self-proclaimed "human-in-the-loop" model. If the very people tasked with refining these systems rely on the tools they’re building, it underscores a troubling circularity that could undermine trust in AI’s reliability and ethical foundations.
Background Context
AI training often involves human annotators paid to guide model behavior, but the field has long grappled with the tension between efficiency and quality. Companies like OpenAI and Anthropic have marketed these laborers as essential for aligning AI with human values, yet the practice of using AI to train AI suggests a deeper reliance on automation than publicly acknowledged.
What Happens Next
Expect increased scrutiny from regulators and watchdog groups over the integrity of AI training pipelines, particularly as transparency demands grow. Companies may face pressure to disclose the extent of AI-assisted training, while workers could push for better safeguards—or unionization—to prevent exploitation of their labor.
Bigger Picture
This episode reflects a broader pattern in tech: the illusion of human oversight masking underlying automation. As AI systems become more integral to critical decision-making, the gap between claimed and actual human involvement raises questions about accountability—and whether the industry is racing ahead without the guardrails it claims to need.
