Japanese scientists use AI to design a hydrogel that heals itself and grips underwater

Researchers in Japan have developed an AI-designed super-adhesive hydrogel that clings firmly to wet surfaces, repairs itself after damage, and stays flexible under demanding conditions. The breakthrough, from scientists at Osaka University and collaborating institutions, combines machine learning with laboratory experiments to identify the ideal chemical composition for a hydrogel that is both highly adhesive and remarkably durable.

Rather than testing thousands of chemical combinations through conventional trial and error, the team’s AI model analysed large datasets to predict which molecular structures would deliver the best overall performance, before the predicted hydrogels were synthesised and validated experimentally. According to the study, published in Nature and titled ‘Data-driven de novo design of super-adhesive hydrogels,’ the researchers “establish an AI-driven materials discovery framework for multifunctional hydrogels.”

The resulting material posted underwater adhesive strength exceeding 1 MPa — among the strongest reported in its class — while maintaining high mechanical toughness and elasticity, and the ability to heal itself rapidly and repeatably after damage. Conventional hydrogels typically struggle to combine strong adhesion, durability and self-healing in a single material, especially underwater, where many synthetic adhesives lose effectiveness.

The innovation could open the door to safer medical adhesives, soft robotic components, wearable electronics and underwater repair technologies, the researchers say, illustrating how AI-driven discovery is beginning to reshape materials science.