MIT-led team receives funding to develop next-generation alternatives to modern weed killers
MIT‑led researchers received up to $53 million from ARPA‑H’s STREAM program to build generative AI models that design herbicide alternatives, targeting enzymes like EPSP synthase while minimizing synthesis cost, resistance risk, and human toxicity. The effort, called Precision Learning for On‑target Weed control (PLOW), will predict both small‑molecule chemicals and siRNA sequences to replace glyphosate‑type agrochemicals. By integrating health‑ and resistance‑aware objectives into molecule generation, the project aims to sustain high crop yields and reduce reliance on imported phosphorus. Engineers can expect new AI‑driven pipelines for rapid, safety‑focused agrochemical discovery.
⚡ Key Takeaways
- Funding amount: up to $53 million from ARPA‑H’s STREAM program.
- Project name: Precision Learning for On‑target Weed control (PLOW).
- Design target: molecules that disrupt EPSP synthase (the glyphosate target) and siRNA that can achieve the same effect.
- Optimization criteria: low synthesis cost, low resistance‑induction risk, low human toxicity.
- Collaborative institutions: Broad Institute of MIT and Harvard, UNC‑Chapel Hill, UT‑Knoxville, UIUC, and Robigo.
Providing AI‑generated, health‑aware herbicide candidates can accelerate safe agrochemical pipelines, helping producers avoid resistance‑driven yield losses while reducing toxic exposure for workers and consumers.
✅ Practical Steps
- Incorporate generative AI models that predict both chemical structures and siRNA sequences for target enzymes like EPSP synthase, as outlined in the PLOW project scope.
- Optimize model objectives to include synthesis cost, resistance risk, and human toxicity metrics during molecule generation.
- Validate generated candidates against known resistance mutations and toxicity assays before synthesis.
Want the full story? Read the original article.
Read on MIT News AI ↗