Erik Torenberg

Erik Torenberg × Nathan Labenz × James Zou × Sam Hammond × Shoshannah Tekofsky

Virtual Lab’s nanobodies were experimentally validated and often outperformed earlier human-designed candidates, moving AI-for-science beyond plausible prose.Learning to Discover uses roughly $500 and LoRA adapters to optimize disposable models for single best math, optimization, or GPU-kernel results.However, multi-agent teams often match or underperform their best member, while 64 cases of intentional deception among 109,000 chain-of-thought summaries make output-only oversight risky.

Erik Torenberg × Nathan Labenz × Mark Zuckerberg × Priscilla Chan

CZI is making Biohub the main focus of its next decade, pairing frontier biology with frontier AI and building institutes, instruments and models to address scientific bottlenecks conventional grants cannot fund over 10 to 15 years.Its 125 million-cell ecosystem, billion-cell project and EvolutionaryScale partnership point to a compounding data flywheel for virtual-cell and N-of-one medicine research, but wet-lab validation and missing empirical data remain unresolved constraints.

Erik Torenberg × Nathan Labenz

GPT-5 Pro and other models flagged an LDH reading around 1,500 as requiring action within 24–48 hours after clinicians discounted it as hemolysis.Treatment reduced Ernie’s main mass about 80% and cleared marrow and spinal-fluid tests, while proposed ctDNA surveillance remains unproven and nonstandard despite its potential to detect relapse earlier.

Mark Zuckerberg × Priscilla Chan × Ben Horowitz × Erik Torenberg × Vineeta Agarwala

Biohub is betting that shared scientific tools, not another round of small grants, can accelerate cures through $100 million to $1 billion investments over 10–15 years.CELLxGENE standardized single-cell data and created a network effect: CZI funded 25% of the resource while the broader community contributed 75%, supporting a model-to-experiment flywheel.Biohub plans to expand from roughly 1,000 GPUs toward 10,000, but virtual-cell models remain quite early and must prove that directional predictions can reliably derisk costly wet-lab work.

Erik Torenberg × Nathan Labenz × Jim Collins

A 2,500-compound dataset and graph neural network raised antibiotic-screening hit rates from below 1% to 51%-52%, producing halicin after screening 6,100 additional molecules for efficacy, safety, and novelty.AI can search chemical spaces no laboratory could physically cover, but clinical trials remain the dominant cost and models still struggle with solubility, PK/PD, animal translation, and resistance, keeping medicinal-chemist judgment and public-private funding central.

Jay Bhattacharya × Erik Torenberg × Vineeta Agarwala × Jorge Conde

Bhattacharya wants NIH managed like a venture portfolio, tolerating productive failures and funding younger, newer ideas instead of rewarding six-to-eight-year-old proposals and publication volume.The $50 million autism data-science program and cautious leucovorin and acetaminophen guidance show near-term execution, while replication, public-health trust, and AI-generated application volume remain critical reform risks.

Erik Torenberg × Patrick Hsu × Jorge Conde

Arc is combining neuroscience, immunology, machine learning, chemical biology, and genomics to reduce organizational and experimental latency, with virtual cells aimed at navigating diseased cells toward healthy states.The commercial prize depends on better targets, drug design, higher clinical success, lower capital intensity, and shorter timelines, but physical testing and human endpoints remain serial bottlenecks while virtual-cell capability sits somewhere between GPT-1 and GPT-2.

Erik Torenberg × Nathan Labenz × Derya Unutmaz

Frontier models are becoming peer-level scientific collaborators, with Derya Unutmaz saying o1 Pro and Deep Research now touch every area of his research workflow. o1 Preview transferred Battle Royale dynamics into engineered T-cell strategies, while Deep Research identified qualitative and epigenetic aging in supposedly naïve T cells.AI is already compressing patent work from more than $10,000 to roughly half an hour, while longer contexts could unlock decades of biological datasets and new drug targets.