Generative AI ( artificial intelligence ) technology has taken huge leaps in recent times. Various AI models are now capable of generating stunning content such as text, images, and even videos by just entering a simple text prompt. One of the largest companies in the world, Google, is also coloring in and opening up other potential in the field of AI.
Recently, Google introduced Genie, an AI model that can turn static images or text prompts into a 2D game ecosystem that can be played directly.
Contents
How to Use Genie, AI Game Maker from Images and Text?
Genie is considered an important leap in the potential applications of AI for game creation and virtual world simulation. Come on, just learn more about Genie from Google in the following article.
1. How does Genie work?
You may be wondering, how can Genie understand the dynamics of a game just from images or text. The answer lies in the learning process. A researcher Tim Rocktäschel, stated that Genie was trained with 200 thousand hours of high-quality 2D video games.
From the video recording, Genie learns the relationship between input given by the player, such as pressing a direction button, with changes that occur on the screen.
This way, Genie can build its own models of how objects should move and react in a game ecosystem. So, when you give Genie a static image or text description, it can manipulate that image to create a live, playable game ecosystem.
2. Potential and benefits of Genie
Genie’s ability to create a gaming ecosystem from simple input opens up very interesting opportunities. Richard Song, revealed the possibility of developing Genie as an infinite generator. This means that Genie can be used to create synthetic data that can then be reused to train other AI models.
In addition, Genie has the potential to be used to create realistic and interactive virtual world simulations. For example, Genie can be trained with videos of moving robot arms, so that it can accurately predict the robot’s movements. This capability will also be useful for developing AI-based game creation technology in the future.
3. Genie’s current limitations
Although Genie is very promising, the technology is still in its early stages and has a number of limitations. One of the main problems is the slow rendering speed. Genie can currently only output one frame per second, far short of the speed required for truly playable games.
Additionally, the resulting ecosystem is sometimes inconsistent in the long term as Genie can only remember up to 16 previous frames.
Another obstacle Genie faces is the possibility of hallucinating and producing unrealistic output, such as characters flying or objects melting into one. This shows that despite its sophistication, Genie still does not fully understand the rules of real-world physics and logic. To realize the vision of an accurate world model, AI like Genie must be trained with even larger and more diverse data.
Even though Google introduced Genie recently, the program is still a research project and not yet available to the public. Google also has not provided further information on when this model will be released.
However, Genie opens up huge potential for using AI for various applications in the future. If this technology can be developed further, then complex and realistic virtual environment simulations could become a reality.
Such capabilities will change many areas of human life, such as in the entertainment industry, urban design or robotics. So, do these kinds of AI developments excite you? Or are you worried about the potential negative impacts?