Why are video games the laboratory of artificial intelligence?

The paradox is fascinating. The video game was created to escape man from reality. Here it becomes a training ground for machines. We remain spectators.

Let's change the subject, but only apparently. On Thursday, Google DeepMind shone a spotlight on the future of video games. It unveiled Project Genie: an artificial intelligence tool capable of turning a sentence or an image into an interactive game world.

Type 'snowy forest with medieval castle'. And the algorithm builds an explorable environment. You upload a drawing. And it becomes a playable level. Not just an image generator, but a world assembler. Several AI models work under the bonnet: one creates the scene, another makes it navigable, yet another handles physics and interactions. A small development studio packed into a single subscription service. The game - because that's what it is for now - lasts one minute. So we are a long way from a commercial outlet. The service is in fact only available to Google AI Ultra users. It is still labelled as 'experimental'. But in the stock market, labels count for less than prospects. How have the financial markets reacted to this 'demo'? The stocks of Roblox, Unity Software and Take-Two Interactive closed the next day with double-digit losses. The fear is clear: if creating virtual worlds becomes as simple as writing a prompt, what happens to the graphics engines, the platforms, the studios that now act as intermediaries? Project Genie is not yet a direct competitor of the big publishers. It is more a laboratory than a factory. But the message has come through loud and clear: artificial intelligence is no longer limited to assisting developers. It is trying to replace entire stages of the traditional production chain. For an industry worth over $180 billion a year globally, it only takes one prototype to shake the odds. In video games, as in space, the real game is played on the terrain of innovation. And those who stand still risk becoming mere background.

Now let's change perspective again. Three developers - Giuseppe Caggese (electronic engineer, class of '64), Emanuele Benedetti, co-founder of Lumen et Umbra, Italy's first text-based MUD (1994), and Marco Orlandi - have created a project still in beta phase called Zero-G. At first glance, it could be mistaken for yet another space video game. In reality, under the shell beats the heart of a persistent simulation based on NASA's LOLA and MOLA topographic data. Here, one does not play at being an astronaut in the playful sense of the word, but navigates inside a 1:1 scale Digital Twin where Newtonian physics dictates the rules and concedes no discounts to anyone.

The most obvious merit of this operation lies in the choice of a single-shard architecture based on Node.js and WebSockets. It is a high-level technical gamble that breaks down the barrier of heavy installation to focus everything on immediate web usability. The value lies not only in exploration, but in the ability to transform scientific data into a tangible experience: moving around in this environment means coming to terms with vectors and inertia, transforming gameplay into an exercise in spatial literacy. The stated ambition is to create a connecting infrastructure for the space economy, a place where universities and companies can meet. If the experiment succeeds in turning into a platform where students experiment with orbital logic and companies establish their virtual headquarters, then Zero-G will not only be a brave game, but the first real digital hub for the aerospace sector. The Italian challenge is launched: it remains to be seen whether the gravity of the market will allow this vision to remain in orbit.