Introduction

In 1997, a renowned grandmaster and world champion in chess, Gary Kasparov, suffered a historical defeat that heralded many ‘game over’ headlines. Playing against a Deep Blue supercomputer, Kasparov deployed every conceivable tactic to corner the machine into a scenario for which it had no sample or matrix.

The result? Seemingly random moves that created a deep game tree and eventually secured victory. The question immediately rose: was it the computational power processing 200 million positions per second? Or was it a sign of an actual thought process? Kasparov himself noted sensing the latter and not just statistical reasoning.

Today we know that Deep Blue had rather trivial reasons for its unconventional play — it did what every software does when pushed to its limit, beyond specification and into territory that neither developers, testers or previous users anticipated: it produced a bug. Deep Blue’s catch-all error routine was to choose a random move.

As new machine vs. world champion contests emerged, experts turned to Go as the next battleground. But ‘the organics’ didn’t last long. AlphaGo defeated Lee Sedol, one of the game’s greats, with advanced machine learning. Then, AlphaZero arrived in 2017. Learning solely from the rules and victory conditions, it trained itself and, in just hours, surpassed centuries of human Go mastery.

This is the state of AI today. We have the tooling at our fingertips, we have the frameworks and the engines to gain insight from game data that surpass the knowledge and skills we can gain from studying the game itself by magnitudes.

The biggest challenge today is accumulating the data. The rules of computer games are notoriously harder to describe than round-based board games such as chess or Go and they are way more complex to generalize.

This is the domain of PlayMind, a data protocol to accumulate structural data for gaming whilst playing and putting incentives for players into its mechanics to collect the necessary data for AI engines to infer the rules and topology of games.

The PlayMind Protocol is about to provide the data at scale to push AI even further on its exponential uptick.

With PlayMind, machines and humans enter a co-creative paradigm, in gaming and beyond.