modern games

modern games


The era when AI is only smart, but why is Game AI still stupid?: Opening up obstacles for the development of Game AI in the current era.

Solve the answer to the problem that baffles gamers of this age. Why has the world evolved so far but the AI ​​in the game is still not advanced?

modern games
modern games

We live in an age where everyone is talking about using AI technology as a selling point or creating a competitive advantage for their products or services. But if you go back 10 to 20 years ago, we’ll find those video games. It is the first type of product that uses artificial intelligence as a selling point for business. Whether it’s an animal that represents the gods in Black & White raises the selling point. Reinforcement learning, AI allows animals to change their behavior according to our play style, or like the early games Far Cry, Crysis, and Halo contained challenging and clever AI as the selling point of the game. Because it opens up a new experience for players by using Game AI that has the lure of the ball. Learn how to dodge bullets or team up instead of appearing to shoot like previous games

“Train your creature and watch it develop its personality according to your action towards it.” One of the selling points in Black and White (2001).

But at present, Game AI has not evolved or changed much from the original. Game AI is no longer an important selling point for giant games, especially in FPS games, we will see more clearly that Game AI intelligence ten years ago. How is it today like that.

What happened to the Game AI development industry that is still in the midst of a trend that every company is talking about about AI?

Let’s talk about 4 reasons that reflect the challenges of developing artificial intelligence for the game today:

1. The complexity of modern game worlds makes game AI development more difficult.

The basic ability that a Game AI must have is the ability to go from point A to point B (path finding) without running into any obstacles. Assistance or pre-scripting is very difficult because the environment in the game can change at any time. Therefore, game developers use techniques such as setting a waypoint or writing a script to inform the AI ​​in advance of where to go, where to dodge, or how to go around and attack according to the player’s position.

But when the game enters the era in dynamic / destructible environments, such as Battlefield, where game environments can change dramatically and change in thousands of ways, the development of Game AI is exponentially more difficult.

Because as the environment becomes more complex, the game AI must also become more complex. This may take more time to develop and is not worth the cost and time wasted. Dynamic/destructible environments make pre-programming a Game AI difficult because humans would not be able to pre-script a Game AI for thousands of possible game scenarios. Interested in playing solo against a Game AI in Skirmish mode as well?

2. Game AI, the smarter it is, the more budget it consumes. The results may not be worth it.

If you have to choose between a smarter game AI and better graphics the computational power is limited.Most developers decide to allocate the processing power of the game machine to the graphics system with more vivid results.Smarter game AI optimization comes with risks that are not worth it.

Because players may not be able to recognize the difference in a more intelligent game AI. In addition, the more complex the game AI, the more difficult it is to detect. The more vulnerable game developers are to errors caused by strange decisions of the game AI, the more game developers focus on using simpler and faster shortcuts, such as the Game AI that follows us in games like Uncharted or God of War, rather than investing in building a game AI that follows the player without interruption, Which may cost a lot of budget and time for innovation and design. Game developers will choose a simpler and faster way. That is, having the character seamlessly teleport to follow us while we’re not looking at the character himself.

modern games
modern games



3- Game adaptive artificial intelligence, the dream of game developers is far from reality.

Adaptive AI or Learning Game AI that will learn the characters in the game and change the behavior and

decisions of the characters according to their style

The gamer used to be a dream for many game developers because Adaptive Game AI will make the user every

gamer with challenges that match the capabilities of the player at any time. No matter how many times the player repeats the game

Adaptive Game AI differs from using the player’s stat system to change the nature of the game AI. For example, in Metal Gear Solid V: The Phantom Pain, if the player takes too many headshots, the enemy in the game will start wearing bulletproof helmets. Put, however, Adaptive Game AI means that Game AI is able to adapt the decision-making strategies of the game. Such as attacking techniques or defensive planning until they are skilled or can beat players according to each style of play

However, Adaptive Game AI is practically no different than machine learning to analyze player patterns and patterns. This takes time and bumps the player repeatedly for ten rounds so the game AI can match the player’s skill. Provided that the player plays the same technique over and over again, which is very unlikely in practice, the development of Adaptive Game AI in video games has gradually declined. Some only have to create an Adaptive Game AI against a Default Game AI due to debugging the game itself.

4. Players don’t want the game AI to be smarter, but they want the game AI to make the game “more fun”.

Designing a game AI to be smart is different from designing a game AI to make a game fun. Exactly the right intelligence of a game AI is important. smart

Very, players are not fun. So the developers chose a more straightforward approach for challenging players. Whether it is a matter of increasing health,

Or increase shields, or increase the attack power of the AI ​​and give the AI ​​game a rhythm that the player can

even anticipate

The player can catch up to the track and win, or add a random factor to let the game AI randomly pick their decision not to be the same

In every game because in the end, how complex do we want the AI ​​game to be?

Finally, although the development of Game AI in the games we play usually faces hurdles due to technology and business reasons. But we are entering a new era of AI game development where the goal is not to keep ordinary players entertained. But the matter

It’s about beating professional gamers to come up with gaming techniques that humans wouldn’t expect, for

example, an intelligence research firm is looking into

Artificial intelligence such as DeepMind for StarCraft games.

Perhaps in the future we may see a game AI that can beat world champions with a game plan that humans have never thought of before.

One Comment

Leave a Reply

Your email address will not be published. Required fields are marked *

Back to top button