Expert questions scalability of single-cell data for AI training, citing lack of counterfactual predictive power
anshulkundaje · x · 2026-08-06
Anshul Kundaje discusses single-cell perturbation data (scPerturb), stating he has not seen convincing evidence that it is the right modality to scale for training AI models with decent counterfactual predictive capability in high-impact downstream applications like in vivo, drug discovery, and clinical trials. He clarifies that scPerturb data is not useless; it is useful for many analyses, but exceptions might be immune cell types and a few other systems.
Related event: Stanford Scholar Questions Efficacy of Single-Cell Perturbation Data for AI(2 posts)→
More from Research
- Challenges in Collecting High-Quality Speech and Egocentric Video Datasets — FaithlessnessWeak199 · 2026-08-06
- NJU Introduces AVE-Compass: A Benchmark for Audio-Video Editing — NJU-LINK · 2026-08-06
- Deep Dive into Kimi K3 Architecture: 2.8T Parameters and LatentMoE Details — AxSaucedo · 2026-08-06
- NBER Paper: 19% of Workers Retroactively Edit Profiles, AI Skills Surge — steverathje2 · 2026-08-06
- Nature Publishes Expanded Codebook of Human Transcription Factor DNA-Binding Specificity — anshulkundaje · 2026-08-06
- Reddit User Discovers New LLM Attack Vector: Non-Instructional Text Prefix Bypasses RLHF — Historical-Cod-2537 · 2026-08-06