Memes are not just jokes—they’re a cultural currency. Our recent study (see references) showed that Large Language Models can outperform humans in average meme quality, but Reddit strongly disagreed on this, claiming they could make better memes than both our participants and the AI. Challenge accepted.

In this follow-up project, we aim to scale up our meme comparison study into the wild. We’re developing a public web platform where users can submit memes, vote on them, and compete with AI models. Let’s see if reddit is right.

You will:

• Design and implement a web-based meme submission and rating platform
• Integrate meme generation from AI models
• Set up voting/ranking mechanics (e.g., pairwise comparisons, Elo ratings)
• Deploy the platform and run a large-scale study with real internet users
• Analyze patterns in meme quality, style, and preference between humans and AIs

You need:

• Solid programming skills in web development
• Interest in LLMs (no need to train models, but you’ll be using them)
• Motivation to work independently on a creative and technically challenging project
• (Bonus) Using a meme in your application email.