Insilico Medicine reported the AI‑generated discovery of a bifunctional PROTAC degrader targeting PKMYT1, with preclinical activity in biomarker‑selected cancer models; the work was detailed in Nature Communications and highlights AI’s role in lead generation and molecule design. Insilico said the Chemistry42 platform produced the D16‑M1P2 degrader and identified contexts—CCNE1 amplification and FBXW7/PPP2R1A alterations—where the mechanism shows synthetic‑lethality promise. Separately, a Nature Computational Science paper introduced SciSciGPT, a framework to improve human–AI collaboration in the science of science, proposing methods to integrate AI into hypothesis generation, experiment planning and reproducibility workflows. Authors E. Shao, Y. Wang and Y. Qian argued for structured human oversight and task‑specific AI agents to accelerate discovery while managing risks. Taken together, the reports signal maturation of AI tools from ideation to actionable preclinical leads and operational frameworks, raising governance, IP and validation questions for R&D teams and translational pipelines.