An AI just invented entirely new chemical structures for drugs that were completely missing from its own training data.
April 24, 2026
Original Paper
Multi-Objective Reinforcement Learning for Generating Covalent Inhibitor Candidates
arXiv · 2604.20019
The Takeaway
This RL pipeline is capable of inventing valid chemical motifs rather than just copying what it has seen. It successfully generated covalent inhibitor candidates, which are a specific and powerful class of drug molecules. Most AI drug discovery models are limited by the examples they were taught, but this system can explore the dark space of chemistry. This means we can use AI to find cures for diseases that current medicine has no answer for. It marks a shift from AI as a database to AI as a true chemical architect. The next generation of medicine will be designed by agents that can think beyond human experience.
From the abstract
Rational design of covalent inhibitors requires simultaneously optimizing multiple properties, such as binding affinity, target selectivity, or electrophilic reactivity. This presents a multi-objective problem not easily addressed by screening alone. Here we present a machine learning pipeline for generating covalent inhibitor candidates using multi-objective reinforcement learning (RL), applied to two targets: epidermal growth factor receptor (EGFR) and acetylcholinesterase (ACHE). A SMILES-bas