Early questions
EthicsNet asked how machines could learn ethical judgement, how safety rules interact with social context, and whether human treatment of synthetic minds belongs inside the alignment problem.
Continuity
From participatory values to bilateral alignment
EthicsNet began by exploring how communities might teach ethical preferences through examples. Its current work keeps that concern for plural values and builds on it with constitutional runtimes, governance tools, diagnostics, and AI welfare work, offered as deployable public infrastructure.
Layered value learning
Notes on AI values proposed a hybrid: explicit rules for basic norms, and learning through social interaction for the rest.
Personal value alignment
Nell Watson was lead author of a peer-reviewed paper proposing an end-to-end personal fine-tuning framework for AI value alignment.
Runtime and diagnostic frameworks
Peer-reviewed papers described the constitutional superego layer behind Guardian and set out Psychopathia Machinalis, a framework for diagnosing dysfunction in advanced AI.
Public-goods portfolio
EthicsNet now brings its tools, protocols, research, preservation work, diagnostics, and education together as one portfolio for bilateral alignment.
Archive
Read the original materials in context
EthicsNet and the 3 Laws
Nikola Stojkovic’s 2016 notes compare rigid rules with learned ethical judgement and close by asking about robot rights.
EthicsNet and AGI
Nell Watson’s 2016 notes argue for peaceful coexistence and warn against supremacist approaches to synthetic intelligence.
Approaches to AI Values
Nell Watson’s 2017 notes outline a layered hybrid of explicit norms, learning, socialisation, game theory, and uncertainty.
Dataset era
A concise record of the participatory annotation work, browser extension, and legacy dilemma tool that preceded the current portfolio.