AI as a catalyst for transforming scientific research: a perspective

Checked 1 Oct 2026. Open license (CC-BY, CC-BY-SA, CC0, public domain): full text indexed and used in synthesis.

Artificial intelligence (AI) is revolutionizing how we conduct, scale, and reimagine scientific research. Unlike prior technologies that amplified human capability within existing paradigms, AI is redefining the very steps of scientific inquiry - from scientific hypothesis generation to experimental validation - and breaking down barriers that have long stymied progress across disciplines. AI has emerged as a transformative tool in scientific research, widely recognized for its contributions to groundbreaking achievements in highly complex domain-specific tasks. Nevertheless, beneath these remarkable successes, systemic vulnerabilities exist that threaten the authenticity of AI-enabled scientific research. Several interconnected challenges are particularly prominent: Large Language Models, now extensively employed for mining data from millions of research papers, face difficulties in extracting reliable information; AI models that learn patterns from training data may generate “hallucinations” that appear valid but are actually false or physically impossible; and both issues are amplified by a persistent lack of high-quality experimental data. Addressing these challenges is not merely a technical necessity, but also a safeguard for the integrity of scientific research.

• Authors across the bundle flag risks of research automation: Lu et al. 2026 warn of taxing overwhelmed review systems and adding noise to the literature, Li et al. 2025 highlight hallucinations that appear valid but are false, and Resnik et al. 2026 enumerate ethical issues including increasing rates of biased, erroneous and deceptive research. Trace →
• Li et al. 2025 explicitly periodize the field, naming Boiko et al. 2023's Coscientist and Huang et al. 2025's Biomni as markers of a 'scientific-agent phase' beginning in 2023 — framing these independently developed systems as milestones of a single transition to autonomous experiment design and workflow iteration. Trace →

• AI for science and agentic research