A year ago, the prevailing story was simple: everyone is using AI for everything. A widely shared analysis on Hacker News takes that narrative apart with hard usage data, and the picture it paints is far messier than the headlines suggest. The reality, according to Hacker News, is closer to “some people are using AI for some things.”
This matters now because companies, investors, and policymakers are making big bets on the assumption that adoption is near-universal and accelerating. The data says otherwise. Let’s clear up the myths one by one.
Myth 1: Once people try AI, they use it for everything
The numbers don’t support it. The Hacker News piece points to Gallup, which found that among Gen Z, the most AI-aware group, adoption has basically stalled. Roughly a third use AI only monthly or every few months. Microsoft’s new US AI Diffusion data, built on anonymized telemetry, reports that just over 30% of the working-age population uses AI meaningfully, defined as at least 90 minutes a month. That’s up only 3 points from late 2025.
Most people who try AI become occasional users, not power users.
Myth 2: Everyone is using AI
Also false. Triangulating across Gallup, Microsoft, Datos, the Searchlight Institute, and The Argument, the analysis lands on a rough three-way split:
- About one third actively use AI
- About one third use it occasionally
- About one third never use it at all
The Datos study backs this up with real-world device data: 62% of desktops visited AI tools zero times in a month. The Argument found most Americans use AI once a week or less. Whichever source you pick, “everyone” is a stretch.
Myth 3: Sentiment is warming as the tools improve
This is the one that should worry the boosters. Usage has barely moved in the past six months to a year. The thing that has changed is negative sentiment, and it’s climbing. Gallup’s Gen Z poll shows anger about AI jumping roughly 40% year over year.
The Searchlight Institute found the top three concerns are job loss (42%), privacy violations (35%), and misinformation (33%). AI carries just a +8% net positive societal rating, sitting right next to social media at +7% and well below cell phones (+68%), the internet (+67%), and solar (+65%). People aren’t ignorant about AI. Many use it and still rate it poorly.
Why the gap exists
The author offers an honest diagnosis: the “everyone uses AI” narrative reflects a bubble around early-adopting knowledge workers, which includes much of the tech press. If your whole feed runs on ChatGPT and Claude all day, universal adoption feels obvious. The telemetry says you’re in a minority.
What stands out here is the regulation signal. A solid majority told Searchlight the government should prioritize safety and privacy rules for AI, even if that slows US development relative to China. That’s a direct challenge to the “move fast or lose the race” framing coming out of many AI labs.
What practitioners and businesses should do
- Stop assuming your users already use AI. If two thirds of the market is occasional or absent, onboarding and education aren’t optional. They’re the product.
- Sell value, not vibes. The Argument concluded people “aren’t really buying the bullish case” CEOs are pitching. Show concrete, individual benefit that outweighs real concerns about jobs, privacy, and accuracy.
- Build for the middle. This isn’t binary. Most people sit on a continuum between heavy use and total avoidance. Design for the cautious occasional user, not the power user who already loves you.
- Plan for regulation. Public appetite for safety and privacy rules is strong. Treat compliance as a near-term reality, not a distant maybe.
The author’s analogy is sharp: telling people how useful AI is sounds a lot like telling them how important protein is. Knowing something is good for you doesn’t mean you change your habits. For a market priced on explosive, universal adoption, that gap between the story and the behavior is the number to watch. You can read the full breakdown at the original source.