Observatory of American Life  ·  CHIP50

Tech
& Society

Six years of national survey data on technology and American life

See the Reports Connect to the data
Mission

Since 2020, CHIP50 and the National Internet Observatory have asked nationally representative panels the same questions, wave after wave — nearly one million survey responses over six years. This site reports what they show about social media, artificial intelligence, and digital communication: who uses these tools, and what changes in American politics, trust, and mental health alongside them.

Reports
REPORT · Platforms, Democracy
Who Posts About Politics?
On every platform, posting about politics is a minority activity. The people who do it look much the same across all twelve tracked: ideologically extreme, educated, male, depressed, conspiratorial.
INTERACTIVE · Mental Health, Platforms
Social Media Intensity and Depression — Interactive Heatmap
A platform-by-frequency view of PHQ-9 depression severity among U.S. adults (Wave 35). Toggle between overall depression and the suicidal-ideation item, and adjust for demographics to see how much of the heavy-use pattern is really about who uses each platform.
REPORT · Mental Health, Platforms
Does Social Media Use Drive Depression?
A PHQ-9 analysis across 21 survey waves finds that demographics explain most of the raw association. A modest link still persists on some platforms, for young women and for heavy users.
REPORT · AI, Demographics
America Meets AI
Weekly AI use jumped from 22% to 29% in a year. The divides that matter are education, age, and income rather than party, and the gender gap ranks among the largest of them. From a 30,000-person national panel.
AI@Work Series 13 Reports · View series →
01
REPORT · AI@Work
Who Uses AI at Work
A majority of U.S. workers now use AI on the job, though adoption divides sharply by education, income, gender, and age. Who uses AI turns out not to be who trusts it.
02
REPORT · AI@Work
What Workers Actually Use AI For
Troubleshooting and brainstorming top the list of workplace AI tasks; writing code does not. The pattern is of a broad thinking aid, and it splits sharply along demographic lines.
03
REPORT · AI@Work
Faster, But Not Freer
Most workers who use AI say it makes them more productive. The time it frees tends to fill with more work, expectations are rising, and workers see the employer as the main beneficiary.
04
REPORT · AI@Work
Don't Ask, Don't Tell
A majority of workers use AI mostly without mentioning it, without a rulebook, and often after feeding it confidential data. Two-thirds of business owners suspect their employees use AI without disclosing it.
05
REPORT · AI@Work
Whose Job Is Next?
A near-majority of workers expect AI to cost people in their field their jobs eventually. Far fewer fear for their own. Anxiety runs highest among the young and among white-collar workers.
06
REPORT · AI@Work
Do Degrees Still Matter in the Age of AI?
The public worries that AI is devaluing college degrees. Hiring managers disagree: they still reward degrees and technical majors, and about half now screen résumés with AI.
07
REPORT · AI@Work
The AI Training Gap
Most workers get no AI training from their employer, and what training exists flows to the already-advantaged. Workers say the obstacle is time rather than willingness, and they are clear about who should pay for it.
08
REPORT · AI@Work
Rein In the Titans, Cushion the Workers
Americans distrust AI and its corporate elite, and they back a worker safety net across party lines. They divide most sharply by AI use itself: the more they use it, the further they move from those who do not.
09
REPORT · AI@Work
Doubling Down on the Diploma
The fear is that AI is making college obsolete. American parents report the opposite: on balance, AI makes them more likely to send their kids, not less.
10
REPORT · AI@Work
Main Street's AI Bet
Most small-business owners who see AI as relevant already use it and plan to spend more. What they buy is subscriptions rather than systems of their own, and the cost savings remain unproven.
11
CAPSTONE · AI@Work
Engagement, Not Fear
Who actually drives demand for an AI safety net? The usual class story points to the fearful. The strongest support comes instead from the engaged.
12
SYNTHESIS · AI@Work
The AI & Work Divide
A synthesis of eleven AI@Work studies: who uses AI, who trusts it, who is protected, and who wants protection all fall along a small number of durable lines. The central axis is engagement versus fear.
13
REPORT · AI@Work
The AI Gender Gap
Across nearly two hundred measures of using, trusting, feeling and fearing AI, men out-engage women almost everywhere. The gap is not about jobs. It is about confidence.
Rf
REFERENCE · AI@Work
The AI Divides — A Reference Note
A shared definition of the demographic divides that structure AI use and attitudes, so every report in the series names, ranks, and frames them consistently.
AI@School Series 8 Reports · View series →
01
REPORT · AI@School
AI Goes to School
Two-thirds of currently enrolled students have used AI for their coursework, and most of them use it lightly. They treat it more as a tutor than a ghostwriter.
02
REPORT · AI@School
I Learn, They Cheat
Students claim AI's benefits for themselves and assign its harms to everyone else. They believe it helps them learn while making their peers learn less, even when the behavior is identical.
03
REPORT · AI@School
The Rules Vacuum
Two-thirds of students use AI for schoolwork, but only one in ten says instructors encourage it, and half say it isn't built into any required courses. AI adoption has outrun the rulebook.
04
REPORT · AI@School
A Leveler With an Asterisk
For students least able to pay for academic help, AI is often the irreplaceable option. Those who stand to gain most from it tend to have the least access.
05
REPORT · AI@School
The Credential Question
Americans think AI is denting the value of a college degree, and the discount they apply is uneven: bachelor's degrees and the humanities are marked down, while STEM and advanced degrees hold or gain.
07
REPORT · AI@School
Curious and Afraid
The generation that has folded AI into its coursework is also the most anxious about it. Concern and fear outrank every positive emotion except curiosity.
08
REPORT · AI@School
Fluent but Credulous
Two-thirds of students rate themselves skilled with AI, and most grasp that it can err. But on the misconceptions that matter when using AI for schoolwork, fluency masks credulity.
09
CAPSTONE · AI@School
Engaged, Uneasy, and Uncritical
Students have adopted AI about as fully as any group in the country, and they are not comfortable with it. Heavier use makes them more excited rather than more careful or more skeptical.
The Data

Every wave repeats the same core questions. That is what makes change measurable rather than inferred.

0
Survey Waves
0
Respondents Per Wave
0
Platforms Tracked
0
AI Analysis Tools
Social Media Platforms
  • Facebook, Instagram, X/Twitter
  • TikTok, YouTube, Snapchat
  • LinkedIn, Reddit, WhatsApp
  • Truth Social, Gab, Parler
  • Mastodon, Bluesky, Threads
  • Usage, frequency, trust, posting
Demographics & Politics
  • Age (8-category), Gender
  • Race / Ethnicity (5-category)
  • Education, Income (10-category)
  • Party ID (3-point and 7-point)
  • Ideology (7-point scale)
  • All 50 states + DC
Health & Attitudes
  • Full PHQ-9 mental health scale
  • Depressive symptom severity
  • Institutional trust (6 items)
  • Trust in science, gov., media
  • Conspiracy beliefs
  • Voting behavior, protest activity
Connect

Ask the panel a question.

You connect an AI assistant to the server and put the question to it in ordinary English. The assistant chooses the analysis tools and runs population-weighted queries against the live database.

What comes back is a table, a trend, a regression with confidence intervals, or a finished PDF report. The privacy rules sit in the query layer rather than with the client: small cells are suppressed, mental-health measures return only as population aggregates, and no personal identifiers leave the enclave.

Once connected, try:

"What variables are available in the CHIP50 dataset? Show me how TikTok usage has changed across all waves."
The server URL
https://chip50-mcp-zbqg33tava-uc.a.run.app/mcp
  • Hand it to your assistantPaste the URL into Claude, ChatGPT, or any MCP-capable client and ask it to connect. It will take you through the rest.
  • Or add it yourselfClaude: Settings → Connectors → Add custom connector. ChatGPT: Settings → Connectors (developer mode). Cursor, VS Code, and other clients: add it to your MCP config.
  • Or use oursTech & Society Insights is our own chat, already connected and open to approved accounts — nothing to install, nothing to configure. chat.techandsociety.ai
  • You need access firstThe server answers only to approved Google accounts. Sign in and it will tell you where you stand.