AI Safety Researcher Linch: Why We Need Simpler Explanations of AI Risks

In an article published in October 2026, AI safety researcher Linch criticizes the current state of communication in the field. He argues that while there are many discussions about AI risks, there is a lack of clear, simple, and public-facing basic explanations. Linch points out that existing explainers often contain too many extraneous details, rely on overly complex analogies, or focus on responding to internal debates, making it difficult for general readers to grasp the core risks.

Using his own experience as an example, Linch notes that when he wrote his "simplest possible case for AI doom" explainer, he found previous materials either sounded like science fiction or were full of insider jargon. He suggests that researchers should try to explain risks using concepts understandable to programmers or other technically minded individuals, rather than assuming everyone shares the same knowledge base. This targeted simplification is not about lowering standards but about increasing accessibility and persuasiveness.

Addressing concerns that "low-quality explanations will drown out good ones," Linch expresses skepticism. He believes reality does a decent job of sorting good arguments from trash, and notes that there are currently very few truly excellent explainers. He emphasizes that if something is sufficiently important, you want "more shots on target," meaning different styles and audiences can complement rather than compete. He calls for researchers in the top 90th percentile of writing ability to spend two weeks writing their own versions to fill this gap.