Collective intelligence for AI-assisted chemical synthesis
A new computational framework called MOSAIC has been introduced to assist chemists in chemical synthesis. It utilizes collective knowledge from millions of reaction protocols and has achieved a 71% success rate in experimental validations. This model not only facilitates the realization of new compounds but also aids in discovering new reaction methodologies.
- ▪MOSAIC is built on the Llama-3.1-8B-Instruct architecture and trains 2,498 specialized chemical experts.
- ▪The framework delivers reproducible and executable experimental protocols with confidence metrics for complex syntheses.
- ▪MOSAIC has successfully realized more than 35 new compounds across various fields, including pharmaceuticals and materials.
Opening excerpt (first ~120 words) tap to expand
Article Published: 19 January 2026 Collective intelligence for AI-assisted chemical synthesis Haote Li ORCID: orcid.org/0000-0002-8146-50661 na1, Sumon Sarkar1 na1, Wenxin Lu ORCID: orcid.org/0000-0001-6388-77431, Patrick O. Loftus ORCID: orcid.org/0009-0009-5952-10901, Tianyin Qiu ORCID: orcid.org/0000-0002-3490-84691, Yu Shee1, Abbigayle E. Cuomo1, John-Paul Webster1, H. Ray Kelly ORCID: orcid.org/0000-0003-3811-06622, Vidhyadhar Manee2, Sanil Sreekumar2, Frederic G. Buono2, Robert H. Crabtree ORCID: orcid.org/0000-0002-6639-87071, Timothy R. Newhouse ORCID: orcid.org/0000-0001-8741-72361 & …Victor S.
…
Excerpt limited to ~120 words for fair-use compliance. The full article is at Nature.
Discussion
0 commentsMore from Nature
