Researchers have been successful in embracing a ‘curiosity’ implant within a recently developed AI model at the University of California. The experiment was conducted over the ‘curious’ model AI and a normal model of AI, where both were applied to play a game of Super-Mario. While the curious system developed ways to escape, the normal model kept hitting the walls with no further plans.
This system can be certainly designated as one of the greatest accomplishments because most of the AI models work upon ‘Reinforcement Learning’ technique in which rewards are provided to the system for accomplishing the task. But it fails in autonomous and responsive situations and thus this implant can certainly be a boon for future Artificial Intelligence Systems.
A machine being curious to learn will certainly help sustain newer methods with better and optimized solutions.
Image Source: livemint