Shriya Biddala
Pronouns: she/her/hers
Research Mentor(s): Eric Gilbert
Research Mentor School/College/Department: / Information
Program:
Authors:
Session: Session 5: 2:40 pm – 3:30 pm
Poster: 7
Abstract
With the rise of generative artificial intelligence models, both corporate entities and open-source communities have intensified their investments in creating and accessing sophisticated models. This trend has sparked a debate regarding which entity, open-source communities or private corporations, innovates faster. Furthermore, there is uncertainty about whether decentralized open-source communities or private corporations make quicker progress. One party may have a lot of money to invest in innovation while another party has more manpower to innovate. To address these questions, I employed a metric called ELO scores, comparing the innovation and progress of various models. These scores were derived from user preferences in a Chatbot Arena, where individuals selected between outputs of two language models (LLMs). Over a one-year period, more than 300,000 preferences were recorded. The results of this analysis highlight the relative innovation and progress rates between different models and contribute to understanding the dynamics between open-source communities and private corporations. Ultimately, this research aims to uncover whether corporations, with their ability to develop and commodify tools, have an advantage over open-source communities, potentially impacting the innovation landscape within the developer community using open-sourced models and peoples lives as we continue to adapt and integrate artificial intelligence into our daily lives.
