Part 1: 34 visions of our future with AI, and the people and institutions who wrote them

Change is coming faster than any of us can absorb, and the anxiety it brings calls for a vision: a plan for who we become as our assumptions about ourselves shift beneath us.

teKnoculture’s AI and Work series studies these visions of the future: 34 texts by 26 people and 12 institutions. Each one is on the Visionary Wheel below, with who wrote it and where to read it yourself.

Preview of the interactive page; opens it

Open the interactive page full screen →

We, as a human family, are impacted so knowing possible strategies, discussing and choosing them is my goal. A reader can see everything, read and judge what we found, and also tell us what we missed.

What this Visionary Wheel is, and what it is not

The wheel reads published visions: manifestos, essays, government plans, security directives, papal and denominational statements. These are texts in which an author or an institution says outright what AI will do to work and what should be done about it. The study calls this the declared layer.

The Visionary Wheel asks every text nine questions, one per spoke:

  1. Work itself: Why does work matter?
  2. The human: What is the human’s role with AI?
  3. Governance: Who decides about AI?
  4. Economic system: What economic system does the text want?
  5. The future: What shape does the future take?
  6. Where the fix lies: Where does the remedy act?
  7. Technology itself: What is AI?
  8. Influences: What older idea or movement does the text continue?
  9. Pride and humility: How does benefit move between people?

To see how every text answers them, open the full Visionary Wheel.

Preliminary work looked at what people say about AI and work online: Who is talking about AI and work, in whose words. That graph is preliminary and will be revised, with clearer instructions on how to read it. The study plans these layers:

  • What people say: social media posts about AI and work.
  • What people ask: questions people type when no one is watching.
  • What people do with the visions: how a vision is taken up, adapted, mocked or acted on.
  • What it is like to live through it: the researcher’s own record of working with AI.

Every layer is coded with the same questions, so the layers can be compared: where the people who publish visions and the people who live with them agree, and where they part. This post covers the first layer only. A later article will detail these asked questions.

The 34 texts, by camp

Group Camps Texts
AI companies 1 10
National intelligence and security 1 6
Religious 6: Buddhist, Catholic, Evangelical, Islamic, Jewish, Latter-day Saints 7
Nation-states 2: Chinese state, US state 3
Eight groups of one camp each 8: AI as a normal technology; defensive acceleration; effective accelerationists and the tech right; feminist; Indigenous; libertarian; power concentration and post-labor; scientists’ warnings 8
Total 18 34

The authors are about 26 named people and about 12 institutions, and the two overlap: a pope speaks for the Catholic Church, an apostle for the Latter-day Saints, a rabbi through his movement’s law committee.

AI companies (10)

National intelligence and security (6)

Religious (7)

Nation-states (3)

One text each (8)

Who is missing

Three things about this list matter for reading the wheel.

Companies and security agencies wrote nearly half of it. AI companies wrote 10 of the 34 texts, and national security agencies 6 more.

Eight camps rest on a single text. For those, one author stands for a whole position: Andreessen for the accelerationists, Thiel for the libertarians, Hinton for the scientists who warn.

Almost every voice speaks from a center of power. Feminist and Indigenous authors write from outside those centers; the rest are chief executives, states, religious authorities and established researchers. Additionally, workers appear only as others describe them. That is the nature of the declared layer, since the people who publish visions are mostly people with platforms. It is also why the study has other layers: workers’ own words belong to what people say, ask and do.

How the texts were chosen

Having researched data science for 16 years, and having used image generators and large language models in my business, I select visionary texts seeking to be inclusive since we all are or will be impacted by AI. Additionally, these visionary texts articulate the relationships between AI technology, human interaction and impact (usually in the form of work), and our social/societal structures.

Some texts were analyzed only in part: the publisher’s sample of The Technological Republic, selected chapters of Situational Awareness and The Intelligence Curse, and three chapters of Feminist AI. The Chinese texts were read in English (the Ministry’s own English version; CSET’s translation). The list will change: each new run adds texts, and this post will be updated with a dated note when it does.

How the texts are analyzed

Each text was coded by an Anthropic AI model, Claude, against a codebook the author designed with the coding model and calibrated against his own blind coding, then checked by the author; every answer rests on a quote that a script confirms word for word in the source. Full disclosure: Anthropic’s chief executive wrote one of the texts above, so the coding LLM has a stake in at least one result. I plan to release our full methodology later because my lived experience as a researcher includes Large Language Models (“AI”).