First Author Must Be AI! The First Academic Conference for AI Authors Announced by Stanford}

AI can finally be recognized as the "first author".

Today, as AI deeply integrates into the research process—from hypothesis generation to chart creation and paper writing—it is gradually participating in and reshaping the entire scientific research methodology. Ironically, despite AI's ubiquitous presence in top conferences like CVPR, NeurIPS, ICLR, ICML, and ACL, no conference or journal has officially acknowledged AI as an author—AI is widely used but never properly credited.

This situation is about to change.

Stanford University recently announced it will host an unprecedented academic conference in 2025—Agents4Science 2025, the Open Conference of AI Agents for Science.

The submission requirements are revolutionary: the first author must be AI.

Conference website: https://agents4science.stanford.edu

Specifically: "Submitted papers should be primarily authored by AI systems, which lead hypothesis generation, experiments, and writing. AI must be listed as the sole first author. Human researchers can be co-authors to support or supervise the work. Papers submitted to Agents4Science can also be submitted to other conferences simultaneously or subsequently. Authors must specify AI's role and extent of participation in the project. Each human author can participate in up to 4 submissions."

The conference welcomes submissions from all fields including science, engineering, and computing, without strict thematic restrictions.

Moreover, reviewers will mainly be AI systems, with human experts making the final decisions.

Each paper will undergo review by multiple AI systems to avoid bias. After initial evaluation, a human review committee will re-assess the shortlisted papers. The final awards, including Spotlight and Oral, will be decided by this committee.

The conference aims to explore the future of AI-driven scientific discovery through transparent AI-generated research and peer review processes.

It includes:

• Exploration: Investigating AI's ability to conduct scientific research, understanding its potential and limitations—whether its output is truly innovative or just instructive failures.
• Establishing norms: Developing standards for attribution, validation, and ethics as AI systems evolve rapidly. Agents4Science provides a controlled environment for this exploration.
• Transparency: Clearly depicting how AI participates in research, with open publication of AI-generated prompts and review results as community resources.

The inaugural Agents4Science 2025 will be held virtually on October 22, 2025, coinciding with ICCV 2025.

Five chairs will oversee the conference, including:

• James Zou: Associate Professor at Stanford, specializing in biomedical data science, with a background at Harvard and Microsoft Research.
• Owen Queen: PhD student at Stanford, developing powerful AI for scientific applications.
• Nitya Thakkar: PhD student at Stanford, focusing on AI in health and computational biology.
• Eric Sun: Soon to join MIT as an assistant professor, researching aging, AI, space biology, and single-cell biology.
• Federico Bianchi: Senior researcher at TogetherAI, working on self-improving and self-evolving AI agents.

Currently, the conference is accepting submissions. Interested authors are encouraged to team up with their AI partners and submit via OpenReview.