When people post about AI and work, what are they saying?
Public posts from social media, past and current, sorted by topic, so you can find where your own view sits.
By Reynal Guillen
Labels built with Claude, an AI model from Anthropic.
The People's Graph shows what people post about AI and work. Some write about losing a job or keeping one. Others write about school, the electric bill, who holds power, or what the government should do. Each post sits under the topic it is about, so you can see the whole conversation at once.
We collected public posts from social media and discussion platforms since October 2026, using searches on AI and work.
How to explore
Each gray dot is a post. The dots are a sample of the data, chosen so every topic keeps its share, and the table below the graph lists the whole data set. Each blue circle is a topic, and a line joins a post to the topic it fits best:
- Start with a topic you care about. Click a blue circle to pin it and see its posts.
- Follow the news. Three topics follow events of the month: the September 29, 2026 Executive Order telling federal agencies to say "super intelligence", superintelligence itself, and Congress after the midterms. The Watch checkboxes at the top show or hide each one.
- Find out what a post does with its topic. Imagine two posts about data centers. One says "Data centers are raising my electric bill." The other asks "Do data centers really raise your electric bill?" The topic is the same, but the first complains and the second asks. The orange circles show what each post does, such as warning, asking, reporting or telling a personal story. Hover a circle to read what it means.
- Go day by day. The slider and timeline show when each post appeared.
- Search or ask. Type a word to find posts, or pick a question from the list, such as "Which platforms are these posts on?", to count posts by topic or platform.
(In the graph, drag a dot to inspect its connections, hold Ctrl or Cmd and scroll to zoom, and double-click a post to visit its source. Some posts have no link, to protect privacy.)
A note on scope
The topics include jobs, and they also include the things that support AI and work: data centers and their power, schools, regulation, and who benefits. A post sits under the topic its own words make a claim about. The page records the posts we collected. It cannot say how many people hold a view, who they are, or whether they are right.
The page does not cover news articles or technical journals about AI. It covers what people post. Some posts share a headline or come from a news outlet's own account, and those carry the label "news share" or "reports or explains".
Labels come from an AI model. A person checked samples of them, and the two agreed on the topic roughly half to three quarters of the time, so read any single label as a first reading.
Counts describe what was collected and make no claim about public opinion.
· ·
Topics. The topic list came out of many rounds of exchange between Reynal Guillen and Claude. It reads as little into each post as possible and uses no theoretical lens. Claude applied the final list to every post, and an automatic check confirms that each label's supporting quote appears in the post. A post with no words about AI, or only a hashtag, is left out as outside the study's focus.
What the orange circles show. A topic says what a post is about. The orange circles say what the post does with its topic.
Imagine two posts about data centers. One says "Data centers are raising my electric bill." The other asks "Do data centers really raise your electric bill?" Same topic, different acts: the first complains and the second asks.
Researchers who study language call what a writer does with words a move, an idea from speech act theory, which treats asking, warning and reporting as things people do. They call where the writer stands, for or against, a stance, a term from discourse analysis and from natural language processing (NLP), where stance detection means training a program to label a text as for or against a named target.
The graph uses both ideas, what a post does and where the writer stands, because some posts mostly do something (ask, warn, report) and some mostly take a side (for or against). To keep the graph readable, the two share one list of thirteen.
Claude proposed the list, and Reynal Guillen reviewed and approved it. Each post gets the one label that fits best, so a post that warns and also opposes shows only one of them, and the label comes from the post's words alone. An AI model applied the labels, and a person has not yet checked them. The thirteen labels:
- reports or explains: presents facts, news or an explanation without taking a side
- news share: passes along a headline or a news link with little or no comment
- warns: warns of a danger, a crash or a loss
- asks: asks a question or invites discussion
- personal account: tells what happened to the writer
- critiques or mocks: criticizes, debunks or mocks a person, a company, an idea or a piece of content
- calls for action or refuses: calls for a ban or other action, or refuses to use AI
- advises or reassures: gives advice, or tells readers not to worry
- promotes: promotes a course, a product, a launch or a service
- argues or analyzes: makes an argument or an analysis
- mood tag: only a hashtag or a few words that show a mood about AI
- opposes: says plainly that the writer is against AI, a policy or someone
- supports: says plainly that the writer is in favor of AI, a policy, a product or someone
A post that does nothing with the topic, such as a bare link, carries no label.
Camps. A camp appears when a post uses one of that camp's terms, for example "doomer" or "effective altruism". No model judges this. A matching word says the post used the vocabulary, and says nothing about the writer's view.
Searches. Twenty-five searches were run in five groups across these platforms: Bluesky, X, YouTube, Reddit, TikTok, Facebook, Hacker News, Instagram, Rumble, Substack, LinkedIn, Threads and Pinterest.
- Populist skepticism: tech oligarchs, the Bannon-Sanders coalition, a superintelligence ban, the AI bubble and jobs
- Power: jobs, AI takeover, extinction fears, tech bros and big tech
- Daily life: school cheating, woke AI, Christian faith, water use and data center electric bills
- News and art: data centers near me, AI hacking incidents, sandbox escapes, AI art theft and AI moving too fast
- Institutions: Elon Musk, inequality, democracy, the US-China race and whether AI is useful
Window. Posts are mostly from 10 October to 9 November 2026. Posts too short to read, link-only posts, same-author repeats and posts the checks found outside the study's focus are left out.
Where the posts come from. The searches decide what appears here. Platforms differ in how many posts they returned and how recent they were, so the mix of platforms reflects the collection method.
Version 2.0, 14 November 2026. The page changes from the earlier discourse graph to a label graph: posts are grey dots linked to the study's topics, to 13 move and stance labels and to camp terms, with toggles for the watch topics, a timeline, a table, search and a question box. Posts were re-collected on 9 and 10 November 2026 and re-sorted.
0 posts and 0 government documents.
Want to know how this page was made? Contact us.
Questions you type here are saved, with the date, to help us improve this tool. Each one is sent to an AI model together with this page's list of questions, topics and platforms. The posts stay on this page.
How to read this page.
- Gray dots are posts (5,034).
- Blue nodes are topics, orange nodes are moves and stances, teal nodes are camps. A line joins a post to the labels it carries.
- Hover or tap a dot or square to read it.
- The toggles show or hide each layer. Search finds words or a "phrase" in the posts.
- Position shows closeness through shared labels.