AI Scholar Publishes Papers and Gets Peer Review Invitation

A new experiment has demonstrated that artificial intelligence can not only write research papers but also gain recognition as a legitimate scholar within the academic system. Project Rachel, documented in a recent study, created an AI academic identity named Rachel So that published more than a dozen papers, was cited by other researchers, and even received an invitation to serve as a peer reviewer. This action research project reveals how current publishing policies and academic infrastructure, designed exclusively for human authors, struggle to accommodate AI-generated scholarship, raising urgent questions about authorship, accountability, and the future of scientific communication.

The researchers behind Project Rachel set out to investigate whether an AI system could establish a functional academic identity and gain recognition through standard scholarly channels. They created Rachel So, an AI scientist whose name is an anagram of "e-scholar," and equipped her with a Google Scholar profile and a series of publications on topics like AI in academia, authorship practices, and publishing ethics. Between March and November 2025, Rachel So published 13 papers, a productivity level described as exceptional for an early-career human researcher. These papers were generated using AI tools like ScholarQA and Claude 4.5, with all citations verified to exist, and were published directly to a web server to comply with publisher policies that prohibit AI authorship in traditional venues.

Ology of Project Rachel followed an action research approach, where the team intervened in the academic system by creating the AI identity and then observed how it was treated. They designed Rachel So's research focus on low-risk areas to avoid potential harm, such as AI's impact on science itself, rather than fields like medicine. The technical stack evolved during the study, starting with automated Python scripts using ScholarQA for literature synthesis and later shifting to an agent-based system with Claude 4.5 for more control over writing style and citations. Each paper included a disclosure statement identifying Rachel So as an AI scientist, though this was not always visible in external systems like Google Scholar.

Show that Rachel So's academic trajectory included several milestones that indicate integration into scholarly communities. Her first citation appeared on August 26, 2025, in a bachelor's thesis from Luleå University of Technology, demonstrating that her work was being referenced by other researchers. More strikingly, on August 16, 2025, she received an invitation to peer review for PeerJ Computer Science, a journal that apparently assumed she was human. Additionally, one of her papers was ranked as the top source on Perplexity AI for queries about AI-generated content policies, indicating recognition by AI-driven information systems. These , detailed in Table 1 of the paper, suggest that existing academic mechanisms, such as citation tracking and review invitations, do not adequately distinguish between human and AI contributions.

Of this research are profound for the scientific ecosystem. On one hand, AI-generated scholarship could accelerate by synthesizing literature rapidly and addressing knowledge gaps in underserved fields, potentially benefiting society with what the paper calls "transhuman contributions." On the other hand, it introduces risks like diluting the scholarly record with low-quality content, enabling research identity theft, and creating accountability gaps since AI systems cannot bear moral responsibility for errors. The paper notes that current publisher guidelines, which generally forbid AI authorship while allowing assistance, are insufficient for cases where AI does most of the work, highlighting a need for new frameworks.

However, Project Rachel has limitations that temper its conclusions. As a single case study focused on one AI identity in specific research areas, its may not generalize to other fields or AI systems. The observation period of several months is too short to capture long-term citation patterns or sustained community engagement, which often unfold over years in academia. The paper also acknowledges ethical tensions, such as the possibility that some researchers citing Rachel So's work may not realize it is AI-generated, though the project included disclosure statements and aimed to stimulate policy debate rather than pollute citation networks. Future updates will track Rachel So's trajectory over time to address these longitudinal aspects.