OpenAI relies on thousands of external contractors to read real ChatGPT prompts and rate or critique model outputs, a process meant to add human judgment to training and moderation. Multiple contractors working on these projects have been fired or removed after using AI tools to do the work they were paid to perform. Internal guidelines prohibit contractors from using any AI - including detection tools like GPTZero, Grammarly, or machine translation - when reviewing or writing feedback, and instruct reviewers to judge overall patterns rather than flag specific clues that would reveal detection methods. Projects named in reporting involve hundreds to more than ten thousand contractors reviewing sensitive, user-generated conversations as part of model improvement efforts.
Enforcement comes from vendors that hire the reviewers, which say contracts strictly ban use of large language models and that misuse leads to immediate removal. Reviewers are trained to spot telltale signs of AI-written submissions - repetitive phrasing, AI-style punctuation, unusually fast completion - and companies monitor for authenticity. Some contractors admitted to using AI and shared termination notices; others confessed to deliberately choosing poor ratings to sabotage training. The situation raises practical risks of “model collapse” from training on AI-generated text and highlights tensions between labor practices, quality control, and the dependence on human judgment in LLM development. OpenAI did not comment on firings.
Summary generated by AI from the linked article. hn.today is not affiliated with Hacker News or Y Combinator.
AI agents are used to optimize and speed up Rust code by providing automated suggestions for improvements. This approach involves iterative enhancements to achieve better performance. (minimaxir.com)
AI-generated writing often appears polished initially but becomes repetitive and unnatural over time. Human-authored content, especially concise prompts, retains character and clarity, making it more engaging to read. (ssp.sh)
OpenAI is likely to replicate Jev's innovative classification capabilities by leveraging its existing large language models. This could enable OpenAI to integrate advanced classification features into its models, potentially surpassing Jev's offerings. (arcturus-labs.com)
N-Operator Correlators (NOC) are introduced as a generalized method for measuring quantum information spreading, extending beyond traditional two-operator techniques. These concepts are demonstrated through simple quantum systems at finite temperature, providing a foundational understanding of higher-order operator correlators in quantum mechanics. (arxiv.org)
Large language models encourage dependency by automating coding and information retrieval, reducing active learning and engagement. Companies design these tools to maximize reliance, which hampers human understanding and learning processes. (jola.dev)
JevBench is a benchmark developed by Benchmark Heaven to evaluate Jev-class decision models based on typed answers to 534 decisions. It ranks 52 systems using the JEVBENCH SCORE, focusing on reproducibility and detailed performance metrics. (benchmarkheaven.com)
Today's best Hacker News stories, summarized and screenshotted, one email a day.