Institutions for
the Intelligent Age
New York City Seven conversations One difficult question

salon series on the other side of AI alignment

How can society flourish
with powerful AI?

Public reading and discussion series about institutional change

Explore Salon 01 Read the premise
01

The opening salon

What must be redesigned to 
receive AI?

Glen Weyl, James Evans, and Chris White’s proposal have a proposal for “reverse alignment”: redesigning institutions so that capable AI enlarges prosperity, agency, truth, and liberty.

Main paper
What Humanity Needs to Flourish in the Next Decade, Noema
Where
SoHo · Exact location shared with registered guests
Registration
Register on Luma
Register on Luma

The premise

Alignment has
two directions.

AI alignment asks how machines can be made compatible with human values.

Reverse alignment asks the reciprocal question: how must our institutions change so AI expands human agency?

proposition

Can reverse alignment become a serious program of institutional change?


8 approaches (and counting..)


  1. 01

    Positive alignment

    asks what a flourishing-oriented AI should do.

  2. 02

    Beneficial deployment

    asks where AI can do useful work now.

  3. 03

    Pro-worker and human-centered AI

    ask which technological trajectories we should encourage.

  4. 04

    Reverse alignment

    asks which institutions society must rebuild.

  5. 05

    Full-stack alignment

    asks how models, companies, markets and governments can be redesigned as one interconnected system.

  6. 06

    Public-interest technology

    asks how technical expertise and institutions can advance public values rather than only private objectives.

  7. 07

    Mission-oriented innovation

    asks how governments and coalitions can direct innovation toward concrete social goals.

  8. 08

    Progress studies

    asks what makes material, scientific, and institutional progress possible, and how it can be accelerated.


12 sociotechnical challenges


  1. 001

    Identity

    Ways to establish that a person, role, or institution is real and accountable without forcing everyone into one universal identity system.

  2. 002

    Privacy

    Rules and technical arrangements that preserve consent, context, and freedom from surveillance as AI makes inference dramatically cheaper.

  3. 003

    Provenance

    Reliable evidence of where data, media, models, and claims came from, what changed them, and who stands behind them.

  4. 004

    Data value

    Institutions for recognizing and sharing the economic value created from people’s data, knowledge, culture, and collective activity.

  5. 005

    Agentic collaboration

    How people and AI agents divide work, delegate authority, coordinate across boundaries, and remain accountable for joint outcomes.

  6. 006

    Communal sensemaking

    Practices that help groups interpret evidence, surface disagreement, and form a usable picture of reality without manufacturing consensus.

  7. 007

    Democracy

    Forms of participation, representation, and deliberation that can use AI without quietly transferring political judgment to platforms.

  8. 008

    Law & liberties

    Rights, due process, appeal, and limits on automated power, especially when states or firms can act faster than people can contest them.

  9. 009

    Workplace

    Organizational designs that preserve responsibility, worker voice, tacit knowledge, and solidarity as teams become fluid and AI-mediated.

  10. 010

    Research

    Methods and institutions that accelerate discovery while protecting verification, intellectual diversity, and the slow work of understanding.

  11. 011

    Education

    Learning and assessment built for a world where producing a plausible answer is cheap but forming judgment remains difficult.

  12. 012

    Labor transition

    Bargaining power, income, mobility, retraining, and social protection for people whose tasks and occupations are reorganized by AI.

season

7 questions

The season moves from the reverse-alignment thesis toward six domains where institutional imagination is overdue. Each salon centers one required paper, supported by a short list of additional readings.

  1. 01

    Orientation · 12 challenges

    What must be redesigned to receive AI?

    Is reverse alignment a useful program of institutional change?

    Tension: Institutional imagination versus technological inevitability.

    Readings TBD

    Main paper

    Weyl, Evans & White · 2026
    What Humanity Needs to Flourish in the Next Decade

    Additional readings

    Dario Amodei · 2026
    Policy on the AI Exponential     Cited in the main paper

    Catalini, Hui & Wu · 2026
    Some Simple Economics of AGI     Cited in the main paper

    Bullock, Hammond & Krier
    AGI, Governments, and Free Societies     Cited in the main paper

  2. 02

    Data value · labor transition

    Who gets paid for intelligence?

    If AI reorganizes value creation at the level of data and tasks, what should replace institutions built around jobs, wages, degrees, and copyright?

    Tension: Distributing AI’s gains versus pricing every human activity as an input.

    Readings TBD

    Main paper

    Arrieta Ibarra et al. · 2018
    Should We Treat Data as Labor?

    Additional readings

    Dario Amodei · 2026
    Policy on the AI Exponential, section 2     Cited in the opening Noema essay

    Astra Taylor · 2018
    The Automation Charade     Counterpoint

    International Labour Organization · 2025
    Generative AI and Jobs     Empirical background

  3. 03

    Agentic collaboration · workplace

    The post-org-chart organization

    What forms of trust are needed when work is performed by shifting teams of people and agents?

    Tension: Flexible collective intelligence versus dissolved employment and responsibility.

    Readings TBD

    Main paper

    Acemoglu, Autor & Johnson · 2023
    Can We Have Pro-Worker AI?

    Additional readings

    Valentine et al. · 2017
    Flash Organizations     Organizational experiment

    Brynjolfsson, Rock & Syverson · 2021
    The Productivity J-Curve     Cited in the opening Noema essay

    Catalini, Hui & Wu · 2026
    Some Simple Economics of AGI     Cited in the opening Noema essay

  4. 04

    Identity · privacy · provenance

    Proof without exposure

    How can people prove that they are real without submitting to universal surveillance?

    Tension: Trustworthy participation versus infrastructure for tracking and exclusion.

    Readings TBD

    Main paper

    Adler et al. · 2024
    Personhood Credentials

    Additional readings

    Renée DiResta · 2026
    We Need a Way to Prove Personhood Online     Cited in the opening Noema essay

    Soliman et al. · 2024
    Meronymous Communication     Noema lineage

    Metropolitansky & Larson · 2026
    VeriTrail     Cited in the opening Noema essay

    Salomé Viljoen · 2021
    A Relational Theory of Data Governance     Counterpoint

  5. 05

    Research · education

    After the essay and the paper

    When generating a plausible answer is cheap, which and knowledge institutions become valuable?

    Tension: accelerated inquiry

    Readings TBD

    Main paper

    Sourati & Evans · 2023
    Accelerating Science with Human-Aware AI

    Additional readings

    Dario Amodei · 2026
    Policy on the AI Exponential, section 3     Cited in the opening Noema essay

    DeSci Foundation
    The DeSci Movement     Cited in the opening Noema essay

    UNESCO · 2023
    Guidance for Generative AI     Institutional response

  6. 06

    Communal sensemaking · democracy

    Can AI help a public think?

    Can AI-mediated deliberation reveal common ground without manufacturing consensus or giving the platform hidden political power?

    Tension: legible public opinion versus hidden power in prompts, clustering, and summary.

    Readings TBD

    Main paper

    Small et al. · 2023
    LLMs for Scalable Deliberation with Polis

    Additional readings

    Jack Henderson · 2025
    The Art of Broad Listening     Cited in the opening Noema essay

    HennyGe Wichers · 2026
    Democracy Needs Friction to Function     Cited in the opening Noema essay

    Sandman & Gregori · 2020
    How Tech Tools Helped Taiwanese Activists Turn a Social Movement Into Real Policy Change     Cited in the opening Noema essay

  7. 07

    Law · liberties · state capacity

    The capable state, constrained

    How can the state gain AI-enabled capacity ?

    Tension: more capable public service versus enforcement.

    Readings TBD

    Main paper

    Engstrom et al. · 2020
    Government by Algorithm

    Additional readings

    Bullock, Hammond & Krier
    AGI, Governments, and Free Societies     Cited in the opening Noema essay

    Dario Amodei · 2026
    Policy on the AI Exponential, section 4     Cited in the opening Noema essay

    Jennifer Pahlka
    Understanding the Cascade of Rigidity     Cited in the opening Noema essay

    Darrell M. West · 2021
    Restore the Office of Technology Assessment     Cited in the opening Noema essay


In New York

1 room shaped for candor, disagreement, and the beginnings of common work.

01

Sabeen
Sidiqui

View LinkedIn

How the evening works

135 minutes.
One paper.
Three turns.


  1. 00:00 25 min
    00

    Arrival

    Drinks and unstructured conversation. 

  2. 00:25 5 min
    01

    Welcome

  3. 00:30 20 min
    02

    Summary explanation

    The argument, evidence, and unresolved tension.

  4. 00:50 15 min
    03

    Understand

    Which assumption holds it up, and where does it fail?

  5. 01:05 15 min
    04

    Apply

    Which institution, community, or practice is ready for the test?

  6. 01:20 15 min
    05

    Build

    What would have to change? Name some experiments

  7. 01:35 40 min
    06

    Socialize

    Promising build tables may continue together.

The small-group rule

Never more than five.

People mix between rounds. A group may stay together if they want to keep building.

Open registration · New York City

Bring a sharp comment
Leave with possibility spaces.

Salon 01

Is society ready to receive AI?

Register on Luma