The most observant of our readers might have already noticed a small little detail about Oasis, a caveat that only the most eagle-eyed OSINT enthusiasts would catch – it’s basically a 1-to-1 copy of Mojang’s Minecraft.

And those readers would be right. Essentially, the “first AI-generated game” is nothing more than blatant plagiarism of everyone’s favorite sandbox, trained on thousands of hours of Minecraft gameplay and recordings of corresponding user actions, which resulted in a nearly identical, but worse in every aspect, “game” with a similar visual style, UI, gameplay mechanics, fonts, visual effects, animations, and so on.

One thing that doesn’t exist in the original Minecraft but is front and center in Oasis is, of course, AI hallucinations. Those who have tried it confirm that the experience is incredibly unstable, with environments often morphing into something else when not in the player’s direct line of sight, making the “first AI-generated game” a proof of concept at best, something that its creators, to their credit, openly admit, describing the current iteration of Oasis as a “technical demo.”

AIs have a big problem with truth and correctness – and human thinking appears to be a big part of that problem. A new generation of AI is now starting to take a much more experimental approach that could catapult machine learning way past humans.

Remember Deepmind’s AlphaGo? It represented a fundamental breakthrough in AI development, because it was one of the first game-playing AIs that took no human instruction and read no rules.

Instead, it used a technique called self-play reinforcement learning to build up its own understanding of the game. Pure trial and error across millions, even billions of virtual games, starting out more or less randomly pulling whatever levers were available, and attempting to learn from the results.