The Sundai Club, a collective of tech enthusiasts and innovators, gathers once a month near MIT with a singular goal: to explore the potential of generative AI in creating innovative tools that address real-world problems. Earlier this year, the group developed a prototype specifically designed for journalists, showcasing the transformative potential of AI in the process.
While generative AI has its critics—concerns over data usage, bias, and environmental impact abound—there is no denying its ability to rapidly prototype useful tools. This was on full display at the Sundai Club, a group supported by Æthos, a Cambridge-based nonprofit that promotes the socially responsible use of AI.
The club’s members come from diverse backgrounds, including students from MIT and Harvard, professional developers, product managers, and even a military officer. Each session begins with brainstorming, followed by a decision on which project to focus on. During a recent hackathon, the group turned its attention to tools that could be of use to journalists.
Several proposals were pitched, including ideas to track political content on TikTok, automatically generate freedom of information requests, and summarise video clips of local court hearings to aid local news coverage. Ultimately, the group settled on a tool aimed at helping journalists identify noteworthy AI research papers on Arxiv, a prominent platform for research preprints.
Using OpenAI’s API, the team created a word embedding—a mathematical representation of words and their meanings—of AI papers on Arxiv. This enabled them to analyse data, uncover relevant research, and explore connections between different fields of study. The end result was a prototype called AI News Hound, designed to assist journalists, researchers, or even venture capitalists in navigating and locating key papers, Reddit discussions, and news articles on specific topics. Though the tool was in its early stages, it demonstrated the potential of using large language models to mine information in novel and useful ways.
Nader Karayanni, a graduate student at MIT and regular contributor to the Sundai Club, expressed his amazement at how quickly the project came together. “The fact that we could build that so quickly is amazing. If you’d told me three years ago that we could do this, I would’ve said it was impossible,” he remarked.
The Sundai Club has developed several other intriguing AI tools, including an AI agent for market research and an AI-generated game explaining hackathons. One particularly creative project even turns research papers into TikToks. Gabriela Torres Vives, one of the club’s organisers, highlighted that while these projects are often playful, they also aim to solve genuine problems. “We prioritise creative problem-solving over traditional engineering methods, and our ideas are designed to meet real-world needs,” she said. Looking ahead, the club is planning to collaborate with companies on more ambitious projects.
While many news organisations are understandably wary of how generative AI could disrupt journalism—through content scraping, auto-generating articles, or diverting traffic—the Sundai Club’s hackathon offered a glimpse of the positive potential AI holds for the industry. Tools like AI News Hound could help journalists filter and analyse vast amounts of data more efficiently.
The innovation emerging from Sundai Club demonstrates how generative AI is still in its infancy but offers significant potential. As the group continues its monthly sessions, the possibilities for AI-driven solutions are expanding rapidly.
Reference: https://www.wired.com/story/sundai-club-generative-ai-hackathon-group/
