Is it truly possible for AI to create a video game?


In the days following the announcement of OpenAI’s GPT-4 model, there has been a growing buzz surrounding the potential of ChatGPT to develop games. While the notion of an advanced AI capable of programming games for real people may sound like something out of science fiction, it is actually not too far-fetched. Game developers and enthusiasts have already been utilizing AI technology in various ways.

Using ChatGPT, developers and AI enthusiasts have successfully recreated classic games such as Pong, Tetris, and Snake. Additionally, they have employed the model to generate code for original games and to generate story ideas and dialogue for text-based role-playing games. A recent example involved someone asking the AI to transform the Game Boy Advance game Pokémon Emerald into a text adventure. Even major game studios are exploring the integration of advanced AI for writing game dialogue.

While some believe that the impact of ChatGPT is significant enough to revolutionize our world, others remain skeptical, questioning the true capabilities of this technology beyond its surface-level allure. It is undeniable that chatbots and AI have been around for some time, and OpenAI’s iteration represents a noteworthy advancement. However, the true extent of its impact has yet to be fully realized, despite the constant discussions about AI on social media and in the news. To gain a better understanding of how ChatGPT functions and its current and potential applications in video games, it is crucial to delve into the underlying technology.

If we were to ask video game enthusiasts about their idealized interactive entertainment experience in the future, they might describe something reminiscent of the software featured in Orson Scott Card’s science fiction masterpiece, Ender’s Game. In Card’s novel, he envisioned a military-grade simulation powered by an advanced and enigmatic artificial intelligence.

Dubbed “The Mind Game,” its primary purpose was to assess the psychological state of young recruits by subjecting them to impossible scenarios, testing their mental resilience in the face of inevitable defeat. The game dynamically generated environments and situations, allowing players to perform any action they could in the real world. It even adapted and responded to the emotional and psychological state of the players, evolving over time. At one point, The Mind Game even drew upon the player’s memories to create game worlds tailored to their past experiences.

While disregarding the darker military applications and the eventual sentience of the software, The Mind Game serves as a starting point to discuss the future of video games and artificial intelligence. Why haven’t games, and the AI used to create them and drive virtual characters, reached this level of sophistication? What tools and technologies do developers still require to achieve this hypothetical fusion of AI and simulated reality?

These are the questions that researchers and game designers are just beginning to explore as recent advancements in AI transition from experimental labs to playable products and practical development tools. Until now, self-learning AI, specifically the deep learning subset of machine learning, which has revolutionized self-driving cars, computer vision, and natural language processing, hasn’t significantly impacted commercial game development. This is despite the fact that some of these AI breakthroughs have been achieved through software that improved itself by playing video games, such as DeepMind’s unbeatable AlphaGo program and OpenAI’s Dota 2 bot capable of defeating professional players.

However, there is a promising horizon where game developers may gain access to these tools and start creating immersive and intelligent games that leverage cutting-edge AI research. This would result in development tools that automate the creation of sophisticated games capable of adapting and responding to player feedback, as well as in-game characters that evolve based on interactions. It may sound like fiction, but it is closer to reality than we might think.

Can artificial intelligence really create a game?

Although ChatGPT cannot run video games itself, it is capable of generating the code required to build a game. AI enthusiast Ammaar Reshi used GPT-4 to generate code for games like Snake and SkyRoads by requesting the HTML, CSS, and JavaScript needed to make them run. While the games didn’t immediately work, Reshi could ask ChatGPT for tweaked code, which it provided along with explanations of the changes. Similarly, AI aficionado Javi Lopez created code for a basic version of Doom using the model.

It’s important to note that ChatGPT doesn’t create the games themselves, but rather the code necessary to develop them. Although these games have simpler code compared to modern video games, they are still functional games. When reproducing an existing game, ChatGPT must have familiarity with that particular game.

ChatGPT appears to have knowledge of a wide range of games, including Pokémon Emerald. It can provide a walkthrough of the game, similar to a text adventure that responds to player prompts. People have also used ChatGPT to create original games, particularly by using it to generate dialogue. For instance, the AI model was integrated into a dating sim called Love in the Classroom. Developers have also sought the AI model’s assistance for generating game ideas and advice. In one instance, a modder used GPT AI to enhance NPC conversations in Mount & Blade 2: Bannerlord, aiming for a more realistic and iterative experience.

Additionally, users have the option to play text adventures within the ChatGPT interface. By providing a starting prompt that establishes the world and its rules, users can engage in solo text-based role-playing experiences, similar to playing Dungeons & Dragons online with an AI Dungeon Master. Beamable CEO Jon Radoff utilized the program to create a prompt for a fantasy adventure, complete with command standards, inventory management, and a map. According to Radoff, ChatGPT was able to enforce rules and constraints in a manner that AI Dungeon, a dedicated AI program for text-based dungeon-crawling games, could not.

Is AI being used by any big game studios?

Credits: CG Spectrum

AI plays a significant role in video games, particularly in controlling non-playable characters (NPCs), both as companions and enemies. NPCs can react to player input, adding depth and realism to gameplay. AI is also utilized in multiplayer games to simulate human-like opponents, providing a challenging experience. Procedural generation, another application of AI, is employed in games like Ubisoft’s Watch Dogs: Legion and Minecraft to dynamically generate game content.

While GPT-4 and ChatGPT are relatively new and may not have been integrated into AAA games yet, major studios are exploring similar AI technologies. NetEase, for instance, plans to include a ChatGPT-like model in its Chinese MMO Nishuihan, allowing players to freely chat with in-game characters. The company intends to use “conversational AI” in more games to enhance new tasks and content variety.

Ubisoft recently unveiled a generative AI tool called Ghostwriter, designed to assist video game writers rather than replace them. Ghostwriter generates initial drafts of NPC dialogue barks, which are short phrases or sounds triggered during specific events in the game. This frees up scriptwriters to focus on polishing the overall narrative. The tool, developed by Ubisoft R&D scientist Ben Swanson, generates dialogue barks based on specific requirements, allowing the writer to choose and edit the responses. While Ubisoft hasn’t disclosed specific projects using Ghostwriter, they emphasize that Swanson is not replacing writers with the tool but supporting them in the production process.

The response to Ghostwriter has been mixed. Some view it as a tool that frees up writers to work on more creative aspects of game development. However, others express concerns about potential job displacement, particularly for entry-level positions responsible for writing barks. The impact of tools like ChatGPT and Ghostwriter on game development labor is still uncertain, and opinions vary across social media platforms.

In summary, AI has already made significant contributions to various aspects of video games, and while tools like ChatGPT and Ghostwriter show promise, the full extent of their impact on the industry is yet to be determined.

AI chatbots are revolutionizing the way businesses engage with their customers, partners, and employees. By leveraging chatbot technology, companies are streamlining workflows, saving time on repetitive inquiries, and allowing employees to focus on higher-value tasks. The digital-native younger generations, who prefer messaging-based communication, are further driving the adoption of chatbots. Moreover, the COVID-19 pandemic has accelerated the acceptance of technology among older generations, prompting them to embrace digital communication tools both personally and professionally.

Overall, there is a positive perception of AI chatbots, leading many companies to integrate this technology to stay relevant and meet customer expectations. According to a Salesforce study, 77% of customers believe that chatbots will have a positive impact on their interactions with companies over the next five years. Consequently, chatbots are becoming a crucial component of customer service automation, transforming the customer experience. Industries such as tourism, financial services, healthcare, wholesale and retail, as well as businesses of all sizes, including small family enterprises and even entire cities, have already implemented chatbots to optimize their online presence, enhance customer satisfaction, and streamline their overall business operations.

Will AI replace game developers?


There is a possibility that AI will eventually have the capability to generate high-quality gaming experiences without human intervention, potentially disrupting certain roles in game development. The technology for this is already available and will continue to advance. However, there are two important considerations to keep in mind. Firstly, it is likely that human developers will still have a role to play in some capacity, as their creative input and expertise may be necessary to shape and refine the AI-generated experiences. Secondly, the timeline for these advancements remains uncertain, and it is difficult to predict when we might see an AI rendition of a complex game like Dark Souls.

It is natural to feel overwhelmed by the potential of generative AI, and concerns may arise regarding its implications. However, it is crucial to maintain an optimistic outlook and recognize that the future of AI is yet to be determined. The impact of these innovations on humanity is still unknown. As responsible members of democratic societies, we have the power to shape the integration of AI with human activities, ensuring that it aligns with our values and benefits society as a whole. By approaching these developments with a positive mindset and actively participating in the decision-making process, we can guide the integration of AI in a manner that maximizes its potential while addressing any concerns.

We recognized the broader impact of AI chatbots in transforming communication between businesses and their customers, with positive attitudes towards their adoption. The potential benefits of AI in improving efficiency, customer experience, and business processes have been observed across different sectors.

However, it is essential to approach AI developments with a balanced perspective. The true future of AI remains uncertain, and concerns about its overwhelming capabilities and potential consequences are valid. Nonetheless, by maintaining an optimistic outlook and actively participating in shaping the integration of AI in democratic societies, we can guide its progression to ensure it aligns with our values and maximizes its potential while addressing any associated challenges.

Ultimately, the journey towards fully harnessing AI’s potential in gaming and other industries requires ongoing exploration, responsible implementation, and an understanding of the evolving relationship between AI and human involvement.

Read more

Recommended For You