Google DeepMind’s artificial intelligence team, alongside researchers at the University of California, Berkeley, has trained AI machines to interact with objects in order to evaluate their properties without any prior awareness of physical laws.

The research project drew inspiration from child development and sought to train AI to mirror human capacity to interact with physical objects and infer properties such as mass, friction, and malleability.

The study, entitled Learning to perform physics experiments via deep reinforcement learning, explained that while recent advances in AI have achieved ‘superhuman performance’ in complex control problems and other processing tasks, the machines still lack a common sense understanding of our physical world – ‘it is not clear that these systems can rival the scientific intuition of even a young child.’

Lead researcher Misha Denil and his team set about various trials in different virtual environments in which the AI was faced with a series of blocks and tasked with assessing their properties.

In the first simulation, called Which is Heavier, the AI was given a set of four blocks which were the same size but varied in mass. The system had to identify which of the blocks was heaviest.

‘Assigning masses randomly… ensures it is not possible to solve this task from vision (or features) alone, since the color and identity of each block imparts no information about the mass in the current episode,’ wrote Denil.

TScreen Shot 2016-11-11 at 11.54.41he AI was rewarded if it correctly determined the heaviest block, and was given negative feedback if it answered incorrectly. Through this reinforcement technique, the AI was able to learn that the only way to obtain information on mass was to interact with the blocks and watch how they responded.

The second environment, Towers, featured five blocks arranged in a tower, with some blocks hidden from vision. This time the AI had to work out how many blocks were used – again working within the reinforcement learning framework. Eventually, the AI was able to learn that it had to interact with and pull apart the construction to decipher the correct number.

Through this research, the team claims that AI is capable of solving problems beyond passive perception, without any prior knowledge of physical properties or the laws of physics.