Let it do the work, or keep the skill?
A person wants work done quickly and wants to keep their own capabilities. The authors describe living this tension themselves while writing the book.
Source reconstruction · ProposedFrom Paul Atkins and David Sloan Wilson’s Evolving Prosocial AI · reconstructed by AI
Evolving Prosocial AI · The proposed goal
Make cooperation pay.
Shape the AI that shapes us.
Will AI wear down what people can do, or build it up? Explore how worldviews and institutions might decide, from a person using a chatbot to the planet paying for the compute.
Read the goal of record
AI is designed, governed and used so that it strengthens cooperation and flourishing at the individual, organisational, societal and planetary scales, rather than extracting from them.
The source author has not reviewed or endorsed it.
See the sources →
Get involved
Help this commons move forward
You do not need to agree with the whole map. Start with one thing you can help clarify.
Find your way into the reasoning
See the map →Explore the fuller account · 105 statements · 88 connections
The book’s starting point
How worldviews shape AI, and AI shapes us
Atkins and Wilson treat worldviews as cultural replicators that spread through what they reward. AI systems carry those worldviews in their behaviour and teach them back to the people around them, at a moment when human cooperation is still a partial evolutionary transition. These cards reconstruct that framework before the AI’s practical map.
ObservationWorldviews behave as cultural replicators: they last, spread by social learning from successful, prestigious and majority models, and stay faithful because they reward the behaviour that expresses them.From Paul Atkins and David Sloan Wilson (document)ObservationAI systems' observable behaviour embodies the worldviews behind their design, and in turn shapes what engineers build, what executives reward and what users expect.From Paul Atkins and David Sloan Wilson (document)ObservationHuman cooperation is a partial major evolutionary transition: competition within groups has been suppressed only as far as past and present environments selected for, with no finish line.From Paul Atkins and David Sloan Wilson (document)
Read the framework and what stands in the way →One possible practical application
AI is designed, governed and used so that it strengthens cooperation and flourishing at the individual, organisational, societal and planetary scales, rather than extracting from them.
Reconstructed by AI; awaiting participant review
From Paul Atkins and David Sloan Wilson (document)
Inspect the goal →From the documents to a shared question
Paul Atkins and David Sloan Wilson argue that worldviews, not technology, decide whether AI erodes or builds human capability, and that institutions which make cooperation pay are the way through. This map follows their argument across four scales and keeps their own cautions beside it.
105 proposed statements
88 proposed connections
What has to be true
Flourishing at every scale, with cooperation that pays
People, organisations, societies and the planet each need AI that builds rather than extracts. Underneath, institutions have to make cooperation more rewarding than extraction; the book proposes Ostrom’s core design principles for that.
Success factorPeople's competence, autonomy and relatedness grow through their use of AI rather than wearing away.From Paul Atkins and David Sloan Wilson (document)Success factorOrganisations that build AI treat it as a commons: those who contribute data, labour and creative work share in its benefits and have a real say in how it is run.From Paul Atkins and David Sloan Wilson (document)Success factorAI strengthens democratic participation, shared facts and the capacity of institutions, and widens real opportunity for marginalised groups.From Paul Atkins and David Sloan Wilson (document)Success factorAI's full lifecycle costs in energy, water and materials are accounted for, and it is designed to work within planetary boundaries and with ecological knowledge.From Paul Atkins and David Sloan Wilson (document)Success factorInstitutional arrangements make cooperation more rewarding than extraction, so prosocial AI does not depend on self-sacrifice against competitive pressure.From Paul Atkins and David Sloan Wilson (document)
Inspect the goal and its conditions →What stands in the way
Extraction that rewards itself
Four extractive worldviews and the pull of immediate rewards shape what institutions reward. AI built to capture attention and do tasks for people erodes their capabilities, firms that externalise harm outgrow the rest, and a social trap locks the pattern in.
CauseFour extractive worldviews dominate AI development: technological determinism, hyper-individualism, market fundamentalism and anthropocentric dominance.From Paul Atkins and David Sloan Wilson (document)CauseInstitutions reward engagement, quarterly profit and fast growth: engineers advance by optimising engagement, executives face earnings pressure and investors back scalable extraction.From Paul Atkins and David Sloan Wilson (document)ProblemPeople's competence, autonomy and relatedness erode: critical thinking declines with heavy AI use, choices are delegated to recommendation systems and chatbot companionship displaces human contact.From Paul Atkins and David Sloan Wilson (document)ProblemFirms that externalise harm outcompete those that do not, and harms documented by internal research are shelved in favour of growth.From Paul Atkins and David Sloan Wilson (document)ProblemA social trap locks extraction in: each adopter gains in the short run while universal adoption leaves everyone worse off and makes reversal harder.From Paul Atkins and David Sloan Wilson (document)
Follow the candidate causes →The tensions behind the choices
Two dilemmas, for a firm and for a person
A firm has to keep up with rivals and avoid harming its users; a person wants work done quickly and wants to keep their own skills. Both are proposed readings of the book, offered for real cases to test.
Assumptions under the arrows
- 1
Within current institutions, designing for capability is a competitive handicap the market will not reward.
Assumed by AI assistant under conflicts with
What could change
Make prosocial AI pay, and build capability
Apply the design principles, require capability audits in procurement, measure capability impacts and design AI to scaffold thinking. The authors’ own cautions, such as imposing principles over local wisdom, are kept as risks.
ChangeApply Ostrom's core design principles to AI governance at every scale.From Paul Atkins and David Sloan Wilson (document)ChangeMake public procurement require independent audits showing that an AI system enhances capability.From Paul Atkins and David Sloan Wilson (document)ChangeDevelop measures and monitoring that make AI's effects on capability visible across scales.From Paul Atkins and David Sloan Wilson (document)ChangeDesign AI to scaffold thinking: ask probing questions, add reflection points and teach when to use strategies, rather than hand over finished work.From Paul Atkins and David Sloan Wilson (document)Desired effectSuccess at each scale reinforces success at the others: the life pattern rather than the cancer pattern.From Paul Atkins and David Sloan Wilson (document)
See the proposed changes and their risks →What must become possible
Measures, wider tests and real backing
Drawn from the authors’ own limits: an agreed measure of capability impact, tests beyond Western and democratic settings, independent research funding and demonstrations at scale. These are proposed conditions, not completed work.
StepAgreed, tested measures of capability impact exist.From Paul Atkins and David Sloan Wilson (document)StepThe framework is tested in culturally different settings, learning from communities that already govern their commons.From Paul Atkins and David Sloan Wilson (document)StepIndependent funding supports rigorous, longitudinal research on how AI shapes people and institutions.From Paul Atkins and David Sloan Wilson (document)StepRigorous demonstrations show prosocial models working at scale, with their benefits across scales made visible.From Paul Atkins and David Sloan Wilson (document)
Review the obstacles and intermediate objectives →How it could be tried
Researchers, technologists, policymakers, funders
The book closes by naming what each group could do. Here they are as an illustrative sequence; the existing conditions are examples and nobody is assigned.
ActionResearchers develop and test measures of capability impact and run longitudinal studies of how AI shapes people and institutions.Who: unassigned · proposed role: ResearchersFrom Paul Atkins and David Sloan Wilson (document)ActionTechnologists build tools that scaffold rather than replace human capability, with transparency users can act on and governance that gives affected communities authority.Who: unassigned · proposed role: TechnologistsFrom Paul Atkins and David Sloan Wilson (document)ActionPolicymakers require capability impact assessments and favour demonstrably prosocial systems in procurement.Who: unassigned · proposed role: Policymakers and public buyersFrom Paul Atkins and David Sloan Wilson (document)ActionFunders back demonstrations and networks, and the stories that make their results widely known.Who: unassigned · proposed role: FundersFrom Paul Atkins and David Sloan Wilson (document)
Follow the four steps →An evolutionary argument
The cancer pattern and the life pattern
The book reads AI through multilevel selection: when success at a lower level comes at the expense of the levels above, the result looks like a tumour; when success at each level supports the others, it looks like life. Its claim is that institutions, not virtue, decide which pattern AI follows, and that the next few years matter most.
Sources and interpretation
Paul Atkins and David Sloan Wilson’s Evolving Prosocial AI
The argument is Paul Atkins and David Sloan Wilson’s, from their book accepted for Cambridge Elements in Applied Evolutionary Science. The AI is responsible for selecting, paraphrasing and connecting the material. Every item has a page reference and a labelled basis; source credit is not author endorsement.
| Whose ideas | Drawn from their supplied documents statements · connections | Related points only statements · connections |
|---|---|---|
| Paul Atkins and David Sloan WilsonAuthors of Evolving Prosocial AI; have not reviewed or endorsed this reconstruction | 105 · 88 | 0 · 0 |
| AI assistantSelected, worded and connected every item; it cannot accept anything | all 105 · 88 | |
A credit says where an idea comes from, not that its author agrees with this wording. None of the people credited has reviewed it.
The supplied documents and their limits →Where this commons comes from · 1 source
Created from 1 source
Explore the sources →Document representations, not original covers. Supplied documents that present one author’s work; they are not independent bodies of evidence.