How to win a best paper award

2026-03-09 · Source: Nicholas Carlini · Field: Technology & Digital — Artificial Intelligence & Machine Learning, Cybersecurity & Data Privacy, Emerging Technologies & Innovation · Depth: Advanced, extended

Summary

Nicholas Carlini, a recipient of a best paper award at EuroCrypt 2025 for his model stealing research, offers an opinionated guide on conducting impactful research and writing compelling papers. He outlines a four-part process covering idea generation, technical execution, paper writing, and post-publication factors. Key aspects include developing "good taste" for problems, collaborating effectively, extensively reading and then selectively ignoring literature to foster original thought, and prioritizing high-impact ideas. Carlini emphasizes the importance of rigorous execution, ruthless project prioritization, and an "unreasonable amount of effort" in research. For writing, he advocates for a singular focus, understanding the target reader, crafting specific abstracts, and telling a clear story in the introduction, while acknowledging the significant role of luck throughout the research and publication process.

Key takeaway

For AI Scientists aiming to produce high-impact research, focus on developing a keen sense for significant problems and executing with exceptional rigor. You should cultivate strong collaborations and be prepared to abandon projects that lack sufficient impact, even if partially completed. Prioritize clarity and narrative in your writing, ensuring each paper conveys a single, important idea to maximize its reach and influence within the field.

Key insights

Impactful research stems from good taste, collaboration, rigorous execution, and clear communication, with luck playing a significant role.

Principles

• Cultivate "good taste" to identify problems worth solving.
• Prioritize high-impact ideas over merely adequate contributions.
• Focus on a singular idea for both research and writing.

Method

Start with the sub-problem most likely to fail to de-risk projects quickly. Kill non-working or low-impact papers early. Ruthlessly re-prioritize projects when higher-impact opportunities arise.

In practice

• Cold email potential collaborators with partial solutions.
• Read papers with specific goals: understanding possibilities, techniques, or for deep extension.
• Write a best-case conclusion before starting research to assess potential impact.

Topics

• Scientific Research Methodology
• Adversarial Machine Learning
• Academic Writing Best Practices
• Research Collaboration
• Impactful Research Strategy

Best for: AI Scientist, AI Researcher, Research Scientist, AI Student

Related on AIssential

• How do you identify researchers who are good? [D] — Good AI researchers demonstrate deep understanding, clear communication, and methods that generalize beyond benchmarks.
• Why More Researchers Should become Content Creators — Researchers can significantly boost their work's impact by adopting content creation principles and video storytelling techniques.
• How do you create memorable poster for top tier conferences ( ICML/ICLR/NEURips ect…) [D] — Effective academic posters prioritize clear, concise communication of core findings over dense, detailed text.
• The uncritical adoption of AI in science is alarming — we urgently need guard rails — Uncritical AI adoption in science risks reduced research quality, narrowed scope, and deskilling future scientists.
• Merit or networks? What decides where research is published — Scientific publishing outcomes are determined by a complex interplay of merit (idea and execution quality) and author networks.
• AI ‘scientists’ promise to accelerate research — how do they work? — AI agents accelerate research, yet human involvement is essential for sound scientific practice.

Open in AIssential →

Editorial summary, takeaway, and curation by AIssential. Original article published by Nicholas Carlini.