← Back
Computation

Decentralized Training

Security RiskGovernance/Ethics Risk
Maintain decentralized control over large neural network training, like SETIatHome/Wikipedia for large language models, to equalize access. However, this also introduces some AI proliferation and control related risks.

Resources (1)

Prime Intellect
Initiative

R&D Gaps (1)

AI Could Be Misused

Computation
The risk of AI being misused—whether through malicious intent or unintended consequences—necessitates robust safeguards and countermeasures.
Computation
52