Info: Zenodo’s user support line is staffed on regular business days between Dec 23 and Jan 5. Response times may be slightly longer than normal.

Published 2023 | Version v1
Publication Open

Towards Community-Driven Generative AI

Description

Position Papers of the 18th Conference on Computer Science and Intelligence Systems / ISBN 978-83-969601-1-5 / Page(s) 43 - 50 / DOI https://doi.org/10.15439/2023f5494 

ABSTRACT

While the emerging market of Generative Artificial Intelligence (AI) is increasingly dominated and controlled by the Tech Giants, there is also a growing interest in open-source AI code and models from smaller companies, research organisations and individual users. They often have valuable data that could be used for training, but their computing resources are limited, while data privacy concerns prevent them from sharing this data for public training. A possible solution to overcome these two issues is to utilise the crowd-souring principles and apply federated learning techniques to build a distributed privacy- preserving architecture for training Generative AI. This paper discusses how these two key enablers, together with some other emerging technologies, can be effectively combined to build a community-driven Generative AI ecosystem, allowing even small actors to participate in the training of Generative AI models by securely contributing their training data. The paper also discusses related non-technical issues, such as the role of the community and intellectual property rights, and outlines further research directions associated with AI moderation. Index Terms—Generative AI, Federated Learning, Crowd- Sourcing, Community, Conceptual Architecture, AI Moderation.
 
AUTHORS

Files

Towards Community-Driven Generative AI.pdf

Files (305.6 kB)

Name Size Download all
md5:77b8ddf44f60cc4e8ef185900422b151
305.6 kB Preview Download