Dear ACM, you're doing AI wrong but you can still get it right (anil.recoil.org)

The ACM Digital Library has recently introduced a controversial feature that generates AI summaries of research papers, sparking backlash within the computer science community. Critics argue that these summaries are often inaccurate and are placed behind a paywall, which denies access to authors who may not have active subscriptions. This move appears contrary to ACM's mission of promoting open knowledge, with suggestions that the summaries could distract from peer-reviewed content. The lack of transparency around the underlying AI technology, the model being used, and the auditing processes raises further concerns about the ethical implications of using AI for scholarly dissemination. This situation highlights a critical moment for the ACM as it navigates the intersection of AI and academic publishing. While AI could enhance accessibility in beneficial ways—such as translating abstract content or creating helpful audio summaries—the current approach is viewed as a regression that undermines the integrity of the library's content. Experts propose that the ACM should focus on fostering genuine scholarly discourse through non-algorithmic means, enhancing access to full-text papers, and implementing systems that better verify the credibility and provenance of research. This pivot is essential not only for preserving the community’s trust but also for leveraging AI technologies to genuinely enrich the academic experience.