Last Thursday, we attended another thought-provoking Spilbar session. The topic was something that has exploded in everyone’s news feeds this year: Generative AI.
Generative AI is a type of AI that can generate content such as images, music, videos, or text. Using deep learning algorithms, it is trained on large datasets of examples which enable it to create new content from what it has learned. Its current and potential use in multiple industries is huge.
When it comes to video games, generative AI can be used for creating procedural content, visuals, dialogue and storylines, and even code.
Developing trust in AI tools
In her talk, Vanessa Volz from modl.ai provided an academic perspective on using AI-based tools to support creative processes in game development. She shared that from her experiences in introducing these tools to game companies, trust is very important. Explaining the processes behind training AI and being able to communicate how AI tools work to support the game dev workflow is vital in building trust.
Generative AI and sparking creativity
Miguel Sicart from ITU had a lighthearted take on generative AI. He spoke of how they should be seen as toys instead of tools. Playing with these toys can kickstart the creative process. Every time you manage to break or trick the system, it generates content that could be a potential inspiration for us.
He shared several hilarious and horrific examples of AI "hallucinations" or instances when the AI did not work as intended. For instance, Chat GPT’s encouraging the chatter to harm themselves or Midjourney’s tendency to create misshapen hands.
 |  |
| photo by unknown redditor |
He concluded that there is nothing to fear from generative AI because even if AI does take over the majority of tasks, the results will most likely be uninteresting, bland and boring, like a “botoxification” of society.
AI-assisted fiction writing
Charlene Putney and Martin Pichlmair gave us an exciting sneak peek at their AI project LAIKA, a tool designed to assist creative writers, game writers, and all other fiction fabricators.
LAIKA is pre-trained with the works of classical authors such as Marcus Aurelius, HP Lovecraft, and Edgar Allan Poe and can write in their tone of voice. It's also possible to train the AI to have your writing style! Once you have selected the "writing brain" to use, LAIKA can finish sentences in the selected writing style.
In this example, we saw how H.P. Lovecraft would have finished the sentences above (AI input in red).
After the panel discussion, we had a blast playtesting a platform game by the creators of the adorable Glyde the Dragon. We also saw some very talented students showcase their project, which used Unreal Engine and AI to generate realistic 3D environments within seconds from text prompts.
It was another fantastic Spilbar experience for Gamucatex and we left feeling hyped about AI's rise in the game development scene. While we need to be aware of the ethics involved in the training and use of generative AI, there's no denying that it is changing the game development landscape and will continue to do so for years to come.
Comments