Benefit Ledger · verified · Materials Science
Lawrence Berkeley’s A-Lab, published 29 November 2023 alongside GNoME, used ML, literature NLP, and robotics to synthesize 41 of 58 targeted solids in 17 days, including materials suggested from DeepMind stability data.
Current score
+0.54
High-throughput autonomous synthesis of predicted solids is major (tier 3). Modest AI attribution (NLP + planners assist humans/robots). Peer review with a later correction. Realization is 17-day campaign products. Medium durability.
An active-learning loop proposed recipes; failed syntheses were analyzed. A January 2026 author correction later updated parts of the record, so some details should be read with the correction. The platform is a research autonomous lab, not a factory.
Provided a subset of computationally stable targets.
Some targets came from DeepMind/Materials Project stability data, not the synthesis policy.
Attribution 0.3000 · Credit share 25% · Google DeepMind
Proposed synthesis recipes from literature-trained models.
NLP and active learning planned experiments; robotics and human infrastructure executed them.
Attribution 0.4000 · Credit share 40% · Lawrence Berkeley National Laboratory
A-Lab realized a majority of selected computational targets during a 17-day autonomous campaign.
outcome · supported
A 2026 author correction revised aspects of the published record.
significance · supported
primary sources
Nature · peer reviewed · 29 Nov 2023 · peer reviewed
Nature · peer reviewed · 19 Jan 2026 · peer reviewed
Secondary domains: Chemistry, Engineering
13 Sep 2026 · 0.00 → 0.54
Initial adjudicated seed score under methodology 0.1.
