BOFFINS HAVE taught an artificial intelligence (AI) neural network to recreate Super Mario Bros by watching other people playing it.
In a paper called "Game Engine Learning from Video," the team from Georgia Institute of Technology explains how the (remarkably accurate, considering) result is based on learning not just how to mimic the game, but also how to mimic the engine powering it - the one designed by Nintendo. When you realise that, you can ignore the glitching and lack of details, because although it's not perfect, by golly it's close.
The neural network is given a dictionary of sprites, along with the basic concepts which are enough for it to analyse and work out the rest.
At that point, it goes frame-by-frame and tries to interpret what it sees as a series of explainable rules and behaviours, which is what the rest of us have been trying to do to the Mario franchise since 1982.
What makes this so amazing is the fact that - to reemphasise - the neural network has no access to the code - it has ‘programmed' a platform game from sight and basic logic. Well, okay, most of a game.
As the paper explains: "Our technique has drawbacks, notably we do not represent player death or level transitions, which makes these key types of mechanics impossible to learn. We hope to address these shortcomings in future work. In addition, we are interested in possible applications of this approach to procedural content generation and full game generation."
Another thing the system can't do is work in 3D. To do that would require a lot more information to be provided to the system than was offered at this stage.
Gaming is becoming an increasingly important part of teaching neural networks on how to behave. From the most famous example - Go - there have been examples of games being used to teach human tactics and emotions, with Elon Musk's OpenAI Gym donating $1bn to help machines learn how to play Pac-Man. Sort of. µ
Sort of. µ
Is restoring from backup really the better than prevention?
Allowed anyone to pinpoint locations visited by customers of SVR Tracking
Hackers gained access to systems using unsecured administrator's account
But Canonical's Mark Shuttleworth doesn't agree