VMTech
Discuss a project →

MIT leads $53 million effort to develop AI-designed herbicide alternatives

MIT leads $53 million effort to develop AI-designed herbicide alternatives

MIT will lead an effort worth up to $53 million to develop next-generation alternatives to modern weed killers through generative artificial intelligence. The funding comes from the U.S. Advanced Research Projects Agency for Health (ARPA-H) under its STREAM program, and the project is called Precision Learning for On-target Weed control, or PLOW.

Christopher Voigt, head of MIT’s Department of Biological Engineering, is the principal investigator. The expected team includes researchers from the Broad Institute of MIT and Harvard, the University of North Carolina at Chapel Hill, the University of Tennessee at Knoxville, the University of Illinois at Urbana-Champaign, and Robigo.

AI models for chemical and RNA candidates

The programme will build generative AI tools to identify alternatives intended to control weeds while taking human health into account. The models are expected to predict either chemical compounds or small interfering RNA, known as siRNA, that disrupt plant enzymes targeted by glyphosate and other herbicides.

Glyphosate interferes with EPSP synthase, an enzyme plants need to produce several critical amino acids. Years of herbicide use have enabled some weeds to acquire genetic mutations that make them resistant, reducing the effectiveness of commonly used treatments and threatening crop yields.

More than 1 billion pounds of herbicide are used in the United States each year. Herbicides help limit crop loss and support regenerative farming practices that protect soil, but frequent use can expose farm workers. Other residents may encounter small amounts through food and drinking water, while the article notes possible associations with cancer, metabolic and endocrine disorders, and other adverse outcomes.

Design, screening and greenhouse validation

The AI work will be led by MIT’s Connor Coley and Regina Barzilay, together with Alexander Tropsha of the UNC Eshelman School of Pharmacy. Their models will optimize proposed candidates for low synthesis cost, a low risk of inducing resistance, and lower risks of harm to people or ecological damage.

The system is expected to generate billions of possible candidates. A high-throughput screening team led by MIT’s Angela Koehler and Jaime Cheah of the Broad Institute will test those libraries for molecules that reach their intended targets. The screening results will then be returned to the computational workflow to improve the models.

After narrowing the candidate pools, a third group will conduct greenhouse studies under different agricultural conditions. Neal Stewart and Scott Lenaghan of the University of Tennessee, Patrick Tranel of the University of Illinois, Tim Willson of UNC, and Robigo founder Andee Wallace will lead that stage. The first greenhouse tests of initial RNAi candidates are expected to finish by the end of the programme’s first year.

Supply-chain and exposure goals

The project also aims to reduce dependence on imported phosphorus, a critical ingredient in glyphosate. In parallel, the researchers plan living microbial systems using engineered bacteria, fungi and srRNA to degrade agrochemicals before they reach food, soil or waterways. The systems would convert chemicals and formulation ingredients into biomass as a potential scalable route to lower environmental and human exposure.

For agricultural businesses, PLOW represents a research pipeline rather than an immediate substitute for existing products: its practical value will depend on whether AI-generated candidates can clear screening and greenhouse evaluation while meeting the project’s cost, resistance and safety objectives.

#generativeai#agriculture#herbicides#biotech
Open analytics
On the site 5 views
min read 4 10.11.2026
Instagram

MIT leads $53 million effort to develop AI-designed herbicide alternatives

Open the post on Instagram ↗