Edging ever closer to artificial general intelligence.
As humans, we take for granted the ability to plan and then to act. To plan means to imagine or construct a plan and then to evaluate it. Algorithms exist to evaluate, just as algorithms exist to execute (ie to act). The ability to imagine or construct plans has, however, challenged programmers. This has especially been true in complex or continuous environments. It has also been true when such planning abilities have need to be generalised across different domains. Pascanu et al have proposed an agent that is able to surmount these challenges.
Link to paper: https://arxiv.org/pdf/1707.06170.pdf
You may also like to browse other AI papers: http://www.thesentientrobot.com/category/ai/ai-academic-papers/