Hidden Feature of DeepSeek AI Can Make Boss Fights a Nightmare in Dark Souls 4, Bloodborne 2

There are only a few things that a single person can be known for, and for Hidetaka Miyazaki, it’s how he is able to create unique video games and punishing level designs for those who seek a challenge. However, he might finally be able to take it a step further with the one AI model everyone is talking about.

Released a few days ago, DeepSeek is another AI model released by Chinese specialists in the field. Not only has it been made for a fraction of the cost when compared to the mainstream options such as OpenAI, but it is also more efficient, which is good for every part of the industry in the long run.

Hidetaka Miyazaki and DeepSeek AI: Time for another unexpected collaboration

The last time we got to see a different side of the director was when he came together with George R.R. Martin to help with the backstory of Elden Ring, a title released in 2022 that has broken all records ever since.

Martin essentially gave Hidetaka Miyazaki a strong foundation to work on; however, when it comes to taking help from the DeepSeek AI, the director might just be able to use the model for something more particular, i.e., boss fights.

From the Dark Souls trilogy to the aforementioned Elden Ring, boss fights have been a major part of the punishing level designs in the game. Using the adaptive intelligence the Chinese AI model has to offer, the studio can finally start making their boss fights seem unpredictable once again.

There’s no doubt that the boss fights in Souls games have hit a plateau as gamers of the modern day just memorize all the move sets and steamroll through the boss fights regardless of their difficulty. Fortunately, DeepSeek might be Miyazaki’s key to changing that.

Technicalities Behind FromSoftware X DeepSeek

What’s about to come usually blows over a lot of readers’ heads; however, it is what makes the recently released AI model different and better from its competitors already out there in the market.

In a nutshell, what makes the Chinese AI model so special is how it is more sparse than the other AIs available. To understand fully, try to think of the AI model’s processor as a human brain. While models like ChatGPT tend to use up all of the neurons available, i.e., 100% of the brain, DeepSeek aims to use just part of the neurons, as required.

This helps save a lot of resources in the long run, which for a studio like FromSoftware can be a big dealbreaker to choose DeepSeek over its resource-intensive competitor such as OpenAI’s ChatGPT.