For P1, my group is creating a game around generative AI/DALL-E. In each round, there is a “guesser” and a “creator.” The creator picks a place, emotion and noun and inputs it into DALL-E. The resulting image is shown to the players and players have to guess what each of the three categories are. They get one point for each correct guess and the round proceeds. Some of the questions I have are:
Are the images created by DALL-E too easy or hard to guess? There seems to be a wide variety in the images created by DALL-E and I’m not sure how it might vary from prompt to prompt. For some images, users may get the prompts right away and it might be boring and for some, it may take too long and be boring. In our prototype, we can test this by finding users and seeing what prompts they generate/the result — I suspect that it could differ from user to user based on their creativity.
Does the game get boring after one or two rounds? Though each round lets a different player be creative and generate these pictures, I’m wondering if the game will get repetitive and boring after a few rounds. Does each round differ enough so that the game is entertaining? To test this, our prototype needs to extend past a few rounds. I suspect that players may get bored and perhaps, we need to add an element of social deduction.
Does this game favor people with experience using DALL-E? We are assuming that background and experience with AI does not really affect performance but it could be that people who have used it before know what prompts or words create more chaotic images. I’m curious if this could translate into the game. To test this, we could test a simple prototype where users generate prompts (rather than playing the whole game) and contrast the complexity of images generated from those with experience and those without. I actually think DALL-E is intuitive enough that it will not matter.
