NVIDIA has introduced a generic model for new models in the field of games, NitroGen, which allows AI to automate the game by reading screens, but it also raises concerns among a large number of players that their use may turn into other forms, undermining the players ‘ fun of connecting.
NVIDIA describes NitroGen as a “Generalist Gaming Activation Model” that uses a large number of video games on the Internet to analyze the responses that people make in different settings, and that allows AI to act on behalf by combining virtual controllers.
NVIDIA collected more than 1,000 games and 40,000 hours of information, of which about 35 % came from the action RPG game, about 20 % was a platform jumping game, and there were more than 100 hours of competition, doves, sports, etc. for the different games, less than 100 more hours and 15 more than 1,000 hours.
At the same time, these data are accompanied by the game controller’s operating records, and NVIDIA collects Xbox 360, PS3-to-now Xbox Series X, PS5 handles. Currently AI can determine the accuracy of keys by more than 90 per cent, and more so by 98 per cent on the latest PS5, Xbox Series X, which is a lower rate of recoil-like correctness of 92 per cent.
The model was designed to assist game and robot developers in their interactive design so that NPC or robots could act more in line with the current situation, such as tanks that know how to walk rather than deliver their heads, and nannys that judge the distance between blood levels and enemy attacks to make up for blood.
Another use is to simplify the quality testing of QAs that can be developed in a game. With a large number of AIs in one breath, one detector can perform dozens of combinations at the same time to find more possible problems in a way that is like mine clearance, and even let the AIs kill each other in a game and find a game that could disrupt balance.
However, the release of the model has also triggered a debate among a number of netizens and players that, since it is automated, there is a high potential for people with a heart to make new “bots” and that there is probably no way to get caught in anti-fraud programs.
The reason for this is that, in addition to the behaviour of the NVIDIA version of the AI model, which is already highly similar to the real person, the way it is judged is to read screen images, while the non-traditional form of intruding memory means that, as long as a camera or a pick-up card is installed, there is no need to decipher the system or the game program at all, the means to play from the outside can be achieved like real people, and the anti-fraud program cannot be caught.
Of course, in terms of NVIDIA and groceries, NitroGen theoretically only follows the logic of “most” players, and the team admits that it is not used to build the superstar rivalry, Dota 2 AI, to the extent that it is unhuman. NitroGen’s ability is largely determined by the ability of a single-channel drama to break through and be a member of the general public.
The real concern, however, is that when AI faces an unreached game, the pass rate is 52% higher than the usual blank model, which means that it has a very good base, and it is really hard to say if it’s accompanied by professional material, longer training and a specialized school, to model a game-breaking experience, making it impossible for the NVIDIA technology to inspire an outlaw.
