Forget retraining - AI can now tweak another AI's weights

Researchers have developed a technique called weight-space programming that allows one artificial intelligence model to directly modify the internal parameters of another. By adjusting these numerical values, the secondary model can improve the target system's performance on specific benchmarks without requiring traditional retraining or fine-tuning. This method offers a faster way to optimize existing AI models for new tasks by manipulating their underlying architecture rather than relying on iterative data training.