Not everyone is using AI for everything
Points and comments are a snapshot, not live.
About one-third of Americans actively use AI, one-third occasionally, one-third never, contradicting claims that everyone uses it.
Multiple data sources show AI adoption in the U.S. has plateaued around 30 percent of the working-age population. Microsoft's 2026 telemetry reports 30 percent using AI tools (ChatGPT, Gemini, Claude, Copilot) for at least 90 minutes monthly. Gallup's Gen Z data shows 81 percent use AI at least rarely, but 31 percent use it only monthly or less, and 19 percent never use it. Datos found 62 percent of desktop devices visited AI tools zero times monthly. Searchlight Institute found 58 percent have tried AI: 30 percent use it regularly, 29 percent infrequently. Usage has barely changed in the past year, while anger about AI jumped 40 percent year-over-year among Gen Z. Top concerns: job displacement (42 percent), privacy violations (35 percent), misinformation (33 percent). AI rates only 8 percent net positive societal impact, compared to 67-68 percent for the internet or cell phones.
What commenters are saying
The dominant sentiment agrees with the article: companies are forcing AI into systems where it makes them worse, slower, and more expensive, driven by executive pressure and AI hype rather than actual utility. Contractors and engineers report being paid to add unnecessary AI to deterministic systems despite knowing it degrades performance. One commenter noted being asked to replace simple shell scripts with agents. A secondary theme emerged: the real value of LLMs is generating code to build deterministic tools, not replacing existing logic. Some pushback: one comment suggests smaller models perform better on logic-following tasks than large ones, and that future hybrid deterministic-nondeterministic experiences may integrate AI invisibly.