GOOGLE has developed Artificial Intelligence (AI) which has an ‘imagination’, proving anything humans are capable of, machines can do too.
The tech giant’s DeepMind project has been working tirelessly on creating AI which matches that of a human mind and the search giant has made a major breakthrough after developing ‘imagination’ for its AlphaGo program.
DeepMind set its machine against a computerised version of the puzzle game Sokoban and an alien invasion game and tasked it with imagining a solution so that it could complete the games.
The British company, which was acquired by Google in 2014, wrote in a blog post: “If our algorithms are to develop equally sophisticated behaviours, they too must have the capability to ‘imagine’ and reason about the future.
“Beyond that they must be able to construct a plan using this knowledge.
“We have seen some tremendous results in this area – particularly in programs like AlphaGo, which use an ‘internal model’ to analyse how actions lead to future outcomes in order to to reason and plan.”
The researchers developed “imagination-augmented agents” (I2As) – a neural network that can extract information to help learn and plan for future scenarios, much in the same way a human brain operates.
The I2As can learn many strategies to help it construct a plan.
Both games the AI played involved reasoning and forward planning; in Sokoban, the ‘agent’ has to push boxes on to targets and in the spaceship game the ‘agent’ has to “stabilise a craft by activating its thrusters a fixed number of times.”
The researchers wrote: “For both tasks, the imagination-augmented agents outperform the imagination-less baselines considerably: they learn with less experience and are able to deal with the imperfections in modelling the environment.”