.png){kind=icon}
.png){kind=icon}
Return
Artificial Intelligence without Digital Computers: Programming Matter at a Molecular Scale
.png){kind=icon}
PRE
AI
DOI：10.1002/aisy.202200157.png)
.png){kind=icon}
Original request for help.png){kind=icon}
Share.png){kind=icon}
Save
Abstract
En 中文
As the carriers and executors of algorithms, digital computers are limited by the density of semiconductors on chips, where the quantum uncertainty is significant at the nanometer scale. Based on the mathematical similarity between chemical pathway networks and artificial neural networks, a new construct to achieve artificial intelligence running on a matter is developed. A general theory is derived followed by an evaluation of its fitting capability and an example where a low-temperature plasma is trained to play the board game Tic-Tac-Toe. The plasma can emit a spectrum representing its next move when we are feeding a gas combination carrying the board information. Finally, the fourth state of matter shows a significantly high winning rate against a random-move player, reflecting its own strategies. This work reveals that any matter, with substantial chemical complexity, can process information based on particle collisions, like a programmable analog computer at the molecular level.
Keywords:
artificial intelligence
chemical pathways
intelligent materials
machine learning
neural networks
Journal
IF:
6.1
Papers:
2.0K
Citations:
8.4K