Below is the list of Decentralized AI verticals

Notes

• These are short- to mid-term and don’t require any fundamental innovations in ML (doesn’t mean innovation won’t happen)
• Most things here don’t require a token but require decentralization to remain unbiased, privacy-preserving, non-manipulative, accessible
• Longer version with all the background reasoning can be found here

Specialized models

What?

AI models tailored to industries, cultures, corporations, domains

Examples

LLMs, diffusion transformers, CV and generative video models, general reasoning models for local languages, cultures and verticals such as healthcare, financial services, etc.

Problems to solve (call for startups)

• Pre-training of small and specialized models
• Data pipeline for synthetic data generation
• Decentralized data collection, labelling, curation, human feedback
• Optimization of hardware, NN architectures

Self-sovereign AI

What?

Locally run models and/or privacy-preserving AI

Examples

Empowering individuals to own and control their personal AI systems to enable free, decentralized, unmanipulated inference