The Good:
AI offers some benefits to video game developers such as improving graphics and implementing adaptive gameplay. AI can generate very complex textures and environments that enhance the realism of gameplay, like foliage or water that reacts realistically to a player moving through it, as well as hair, a notoriously difficult aspect of animation. Additionally, AI can be integrated into non-playable character (NPC) behaviors in order to create a truly personalized player experience. Adaptive features of AI NPCs include enemy learning, which would allow enemies to learn from player fighting tactics and adapt accordingly for a more difficult, realistic battle, friendly tailoring, which would make friendly NPCs more helpful to the player, and storyline tailoring, in which the game would learn what aspects of the plot interest the player most and then tailor the plot to be as engaging as possible for any given player.
A prominent example of AI NPC use to enhance gameplay is Rockstar’s leak that Grand Theft Auto VI will include NPCs armed with AI in order to make them more interesting and give them new abilities such as being able to react to the player's smell. An added bonus of these features is that AI would be able to balance difficulty with entertainment, making the game just hard enough to keep players motivated but not too hard that they become discouraged and stop playing.The Bad:
The drawbacks to implementing AI, however, are numerous. On the industry side, although large studios may be looking to cut costs by replacing salaried designers and writers with algorithms, the cost of upkeep for video game AI is possibly more expensive when the requisite highly skilled labor and pricey soft- and hardware are taken into account. Aside from this, it’s an ethical question of whether or not to put developers out of jobs, especially considering the human element that would be lost in doing so. Much like the way that AI art and writing always feels a little ‘off,’ AI video games will fall into the same trap. Artistic generative AI is trained on existing materials, making it incapable of being truly innovative and, even though it took human arts and can emulate it, it will never possess the human context that goes into the creative process and therefore, by my definition, cannot actually create art of any kind. Video game enthusiasts look for and appreciate the human element that elevates the story of a game, and studios are mistaken if they think that no one will notice when it’s gone.AI learning in video games also involve vast amounts of private data collection. The algorithm, in order to create a truly personalized gaming experience, collects player data and synthesizes it into a personality to be catered to, meaning that player data profiles are being stored and could be used with questionable intentions or sold to third parties. AI algorithms, because of the way that they’re built, are also susceptible to new kinds of cheat programs that would threaten the integrity of the game.
Lastly, although computers generally imply objectivity and true randomness, they are still built by people, who do not. AI programs written by people who hold explicit or even implicit biases against certain groups are extremely likely to carry traces of those biases. We’ve seen this happen in recruitment scenarios, where applicants of a certain race or gender are preferred by the algorithm, and in generative art AI, which has been seen to be biased in its training and, therefore, in its depictions of Black people as opposed to white people. In video games, AI could perpetuate harmful stereotypes in similar ways, making gameplay a possibly destructive experience.
AI depictions of Black women smiling from Google Colab, 2020 |
The Ugly:
Generative AI seems to aim to replace, or at the very least devalue, the important work of artists, including screenwriters, digital artists, video game designers, and other professions that rely on the unique artistic talents of creatives, making it an ethical question and even a legal one. The training of generative AI on unlicensed content scraped from the web and then emulated without credit to artists is an intellectual property issue and could be a copyright violation. Midjourney, DeviantArt, Stability, and Runway AI are actually in the middle of a lawsuit right now regarding this issue: some artists are alleging that the AI art programs used art online for training without consent from the artists and arguing that artists must be credited for use of their art, even by AI. A statement from the artists demonstrates the strong sentiment fueling their legal battle: “Though Defendants like to describe their AI image products in lofty terms, the reality is grubbier and nastier. AI image products are primarily valued as copyright-laundering devices, promising customers the benefits of art without the costs of artists.”
No comments:
Post a Comment