§ I — The Thesis
Common sense is not good sense
Common sense has been the white whale of artificial intelligence for seventy years — the tacit grasp of how the world hangs together that no amount of data seems to confer. Our wager is more unsettling: that even were it caught, it would not be enough. At least three faculties travel under the one name. There is consensus, the average of what a community believes. There is the vast stock of unspoken world-knowledge whose absence makes machines brittle. And there is the oldest meaning of all — the power of judging well. Only the last is good sense.
The distinction is not ours; it is the better part of the Western tradition speaking in several tongues. Descartes opens his Discourse with le bon sens, "the best distributed thing in the world," and means by it precisely the faculty of distinguishing the true from the false — a faculty so universally presumed that no one thinks to cultivate it. Kant's sensus communis is not vulgar opinion but a community sense that judges from the standpoint of everyone else. Gramsci, writing in prison, prises apart senso comune — the uncritical, inherited sediment of belief — from buon senso, its small, sound, critical kernel.§0
A system may hold an encyclopaedia of facts, and faithfully mirror the average of human opinion, and still misjudge.
For sound judgement, as David Hume saw, is not the reproduction of what is widely believed or widely done. It is corrected sentiment, generalised from a common point of view: a verdict reached by discounting one's own position and asking how a character or an act bears on those it touches. Consensus is a head-count; good sense is that head-count refined. A model trained to echo aggregate preference has learned the first and not the second — and the gap between them is the whole of ethics.
This gap has become urgent because artificial systems no longer merely answer. They classify, recommend, and decide; and in doing so they reach into the social world and remake it — fixing who counts as creditworthy, employable, or suspect — within an informational environment that those same systems are quietly re-ontologising. The loop between a society and its machines can turn virtuous or vicious. Which way it turns is a question of judgement. This review names that judgement good sense, and proposes a programme in three prepositions for building it, exercising it, and holding our systems to it.
§ II — The Framework
Good sense for, by, and to AI
Three prepositions, three modes of the same concern. The first looks forward to design; the second resides in the act; the third looks back upon consequence. Together they close a circle.
Building good sense in
A design desideratum for any system we are willing to call AI. To engineer good sense is not to hard-code a fixed table of values but to build in the corrective process itself — the capacity to generalise and de-bias sentiment toward a common point of view — together with the explicability to show its working and an attention to what the system anchors, not merely what it outputs.
The system as it judges
An AI running upon a good-sense lens: describing and acting on the world with corrected, context-sensitive judgement. This remains coherent even for what Floridi calls agency without intelligence. To exercise good sense is not to be conscious; it is to have one's agency structured so as to track corrected sentiment and to refuse avoidable harm.
The standard we apply
Good sense as an external measure applied to systems never designed with it — but to which, given their effects, it would have applied. The retrospective, patient-oriented question: did this system corrupt the shared informational environment, re-anchor a social kind perniciously, or set a vicious circle turning? Consequence answers to judgement even where design did not.
§ III — The Signature
One loop, two fates
Floridi's deepest lesson is that a society and the technologies it builds shape one another without end. Design conditions what a system does; what it does has consequences; our judgement of those consequences flows back into the next design. The circle never stops turning.
Nothing in the mechanism decides its direction. The same loop that homogenises taste, narrows the visible world, and trains itself on the preferences it has manufactured can, under a different standpoint, widen sympathy and repair the commons. Good sense is simply what makes the circle virtuous rather than vicious — at every pass.
Hover or tap a station to follow the loop.
§ IV — The Foundations
Three thinkers, one division of labour
Good sense is not invented here from nothing. We borrow its content, its mechanism, and its stage from three bodies of work, each answering a different question.
Hume
David Hume · 1711–1776
Reason, for Hume, is "the slave of the passions": it finds means but cannot set ends. Morality is felt before it is judged, and no ought follows from a bare is. Yet sentiment, corrected from a general point of view, yields a standard that is shared without being a mere head-count.
Good sense is corrected sentiment — intersubjective, not statistical.Epstein
Brian Epstein · The Ant Trap, 2015
Social facts are grounded by what makes them obtain in a given case, and the conditions for grounding are themselves anchored by deeper facts. AI grounds particular verdicts — and, more consequentially, re-anchors the very kinds: what now counts as a risk, a qualification, a person of interest.
Good sense must reach the anchoring of social kinds, not only outputs.Floridi
Luciano Floridi · The Fourth Revolution, 2014
We live as inforgs within an infosphere that information technologies do not merely furnish but re-ontologise. After Copernicus, Darwin, and Freud, a fourth revolution unseats us from the centre of the world of information — and binds us in a recursive circle with our own creations.
Good sense is what turns that circle virtuous rather than vicious.Hume supplies what good sense is; Epstein, where it must intervene; Floridi, the medium in which it acts and the dynamics it must govern. The full argument — with its tensions honestly stated — is set out in the Manifesto.
§ V — The Collegium
An interdisciplinary collective
No single discipline owns judgement. GoodSense.ai convenes five, each contributing a faculty the others lack. We are a working collective rather than a roster of names — and the table has empty chairs.
Artificial Intelligence
What can be specified, learned, and left to judgement — and where alignment by content reaches its limit.
Philosophy & Ethics
The normative content of good sense, the is–ought gap, and the standards by which consequences are weighed.
Social Ontology
How systems become embedded in social reality, and how they re-anchor the kinds through which we sort the world.
Psychology
Sentiment, sympathy, and habit as they actually operate — the cognitive grain that any account of judgement must respect.
Information Technology
The infrastructure of the infosphere: where design choices harden into the conditions of everyday life.
We envision a future in which artificial intelligence is designed and used in ways that reflect our shared humanity and support the greater good. This site is a hub for that work — for discussion, collaboration, and education on the landscape of AI ethics, and for the claim that good sense in AI is becoming not a luxury but a condition of flourishing.
¶ A Call for Participation
Join the inquiry
We welcome researchers, practitioners, and critics across AI, philosophy, social ontology, psychology, and information technology — and the curious from any field who suspect that judgement, not merely capability, is the question of the age. Write to propose a collaboration, contribute to the review, or simply to think aloud with us.