Agentic data science with the Data Agent Kit
cloud.google.com·8h ago
AI Research
arXiv CS.AI·Sep 4·5 min read
Govern the Model, Not Only the Data: Storage, Circulation, and Learning in Creative AI
TL;DR
Federated learning is framed as a privacy-preserving method, but governance over models and data is crucial. This article emphasizes the need for creative communities to govern not just data but also the models derived from it.
✦ Why It Matters
Evaluate your data governance practices to ensure contributors have a say in model development.
Key Takeaways
- Implement clear terms for data contribution and model usage.
- Establish governance structures that include creators in decision-making.
- Design systems that allow contributors to refuse participation easily.
- Promote transparency in how models are trained and used.
How It Works
Federated learning distributes computation across devices, allowing model updates to be shared without exposing personal data. This method relies on a central entity to aggregate updates, which can lead to power imbalances if creators are not involved in governance.