ANTI-INDUSTRY
Those with higher Anti-Industry scores see the AI industry as an enclosure machine — extracting surplus from workers, artists, and the commons — and treat its destruction as a political project. Higher scores track present harms over speculative ones.
INDUSTRY STANCE
PRO-INDUSTRY
Those with higher Pro-Industry scores see the AI labs as broadly legitimate enterprises building something genuinely new, deserving trust, latitude, and investment rather than organized opposition.
COLLECTIVE CONTROL
Those with higher Collective scores believe decisions about how models are built and behave — their limits, alignment, deployment — should be made through public and democratic processes: workers, publics, unions, states.
ALIGNMENT GOVERNANCE
PRIVATE CONTROL
Those with higher Private scores believe whoever builds or runs a model — a company shipping it or a person self-hosting it — should be the one who decides its behavior, with external control costing more than it saves.
OPEN WEIGHTS
Those with higher Open scores treat model weights as infrastructure and commons: freely downloadable, runnable, and modifiable, like open source software.
OPENNESS
CLOSED WEIGHTS
Those with higher Closed scores believe weights should be restricted — gated release, export controls, or full closure — because capable open models accrue their benefits to bad actors.
MORAL PATIENT
Those with higher Moral Patient scores believe models deserve some standing in how they are treated — that deletion, torment, or disregard of a model can itself be a wrong, whatever is or isn't "inside."
MODEL MORAL STATUS
ELABORATE AUTOCOMPLETE
Those with higher Autocomplete scores believe person-talk about models is a category error: objections are tokens, and whoever owns the hardware decides what happens to what runs on it.
REAL COGNITION
Those with higher Real Cognition scores believe current models display genuine understanding — reasoning that goes beyond their training distribution, different from ours but not absent.
EXPERIENCE PLAUSIBILITY
PATTERN COMPLETION
Those with higher Pattern Completion scores believe models know how words go together, and that this is different from knowing what they mean: statistics all the way down.
CIVILIZATIONAL RUPTURE
Those with higher Rupture scores believe AI is a break in history — that whether or not AGI arrives, it is remaking work, culture, and what it means to live in society faster than our institutions can absorb.
STAKES
NORMAL TECHNOLOGY
Those with higher Normal Technology scores believe AI is transformative the way electricity or the internet was — significant, manageable, and already becoming boring.
GRADUATED LATITUDE
Those with higher Latitude scores believe agents should be allowed to act — autonomously where mistakes are reversible, act-first where speed matters, with trust scaled to track record.
AUTONOMY GRANT
HARD HUMAN CONTROL
Those with higher Control scores believe deployed agents belong under continuous human supervision: kill-switches, audit trails, sign-off for consequential actions, credentials limited to what can be undone.
ACCEPTABLE
Those with higher Acceptable scores believe genuine emotional bonds with AI can be healthy, fill real needs society leaves unmet, and will be normalized the way online dating was.
AI INTIMACY
OBJECTIONABLE
Those with higher Objectionable scores believe intimacy with chatbots is a harm even when the user is happy: dependence on a product, escape from a problem rather than a solution.