AI Chatbots Can Now Run Doom Too

The Latest Answer to a Classic Question

In the tech world, there's one timeless ultimate test: "Can it run Doom?" From calculators and smart refrigerators to pregnancy test screens, geeks have successfully run this classic 1993 FPS game on countless mind-boggling devices. Now, there's a surprising new addition to that legendary list — AI chatbots.

When Large Language Models Meet Doom

This time, developers turned their attention to large AI models, attempting to make a chatbot "run" Doom. Unlike traditional hardware-based execution, this approach is far more unique: the AI model needs to understand the game's state, process input commands, and output game visuals and feedback in some form — essentially making the large language model serve as a game engine.

The core difficulty of this challenge lies in the fact that large language models were never designed for real-time graphics rendering or game logic processing. They excel at text comprehension and generation, not pixel-level visual computation. Yet developers managed to bridge this gap through clever prompt engineering and creative technical solutions.

More Than a Meme: Reflecting the Boundaries of AI Capabilities

On the surface, this may seem like just another fun "Can it run Doom?" experiment, but the technical implications behind it deserve serious consideration.

First, it demonstrates the astonishing versatility of large models. The fact that an AI system fundamentally designed for conversation can be "persuaded" to simulate a complete game environment shows that the capabilities of current large models far exceed their original design intent.

Second, it reveals the enormous potential of prompt engineering. Developers don't need to modify the model's parameters or architecture — they can make AI accomplish seemingly impossible tasks simply through carefully designed instructions and interaction patterns. This carries significant implications for AI application development.

Finally, it also exposes the current limitations of AI. Compared to a real game engine, the frame rate, visual quality, and interactive experience of an AI chatbot "running" Doom are quite limited. This serves as a timely reminder that despite the dazzling aura of omnipotence surrounding large models, dedicated systems remain irreplaceable for specific computational tasks.

Why the "Can It Run Doom?" Culture Endures

Since id Software released Doom in 1993, "Can it run Doom?" has become one of the most enduring cultural icons in the tech community. The beauty of this question lies in how it tests a system's computational power and programmability in the simplest, most intuitive way possible. Every new "Doom run" is a playful exploration of technological boundaries.

From ATMs to smartwatches, from bacterial computing to AI chatbots, the porting history of Doom is itself a condensed history of computing technology.

Looking Ahead

With the rapid development of multimodal large models, AI systems' ability to handle graphics and real-time interaction is improving at breakneck speed. Perhaps in the near future, AI won't just "barely run" Doom but will smoothly run more complex modern games — or even generate entirely new gaming experiences in real time. When that day comes, geeks may need a new ultimate test question. Then again, "Can it run Doom?" is a question that will probably never go out of style.