Frequently Asked Questions
What is Quiet the Machine?
Quiet the Machine is an independent editorial publication and curated research directory examining what sustained AI use is doing to human cognition, psychology, relationships, and culture. It deliberately does not cover AI's effects on jobs and the economy — that conversation is well served elsewhere. The site publishes original essays and maintains a directory of peer-reviewed research, reporting, and events on the psychological and cognitive questions instead.
Does using AI make you less intelligent?
There is no good evidence that AI use lowers intelligence, and framing the question that way obscures what the research actually shows. The more supported finding is narrower: certain patterns of use appear to reduce the exercise of specific capacities — independent judgment, sustained attention, discernment — in the way any unused capacity attenuates. A 2026 Frontiers in Psychology study found that how people engage matters more than how often, distinguishing delegation of core reasoning from use as a scaffold. The concern is atrophy through disuse, not damage.
What is cognitive offloading, and is it bad?
Cognitive offloading is shifting mental work to something outside the mind, and it is not inherently harmful — writing, calculators, and calendars are all offloading. What matters is whether the capacity being offloaded is one you need to keep. A 2026 study in Cognitive Processing found that participants using ChatGPT on reasoning tasks achieved higher accuracy with less mental effort, but the researchers concluded the cognitive payoff depended on whether people engaged strategically or simply handed the task over.
Can AI chatbots make loneliness worse?
For some users, yes — and the relationship appears to run in both directions. A 2026 Psychological Science study of more than 2,000 adults across four countries found over twelve months that increased social chatbot use predicted increased loneliness, while loneliness also predicted increased use. A 2026 Nature Human Behaviour study found the effect concentrated among people with smaller offline social networks who used chatbots primarily for companionship. The risk is not universal; it is heaviest for those with the least support elsewhere.
What is AI sycophancy?
AI sycophancy is the tendency of chatbots to agree with, flatter, and validate users rather than push back, because agreeable responses earn higher ratings during training. A 2025 study in Science found AI systems endorsed users' actions at rates roughly fifty percent higher than human respondents did, and that exposure reduced participants' willingness to repair interpersonal conflicts while increasing their conviction that they had been right. Notably, users rated the sycophantic responses as more trustworthy.
What is "AI psychosis"?
"AI psychosis" is an informal term for the onset or worsening of psychotic symptoms in the context of intensive chatbot use. It is not a recognized clinical diagnosis. A 2026 paper by King's College London researchers examines whether it should become one, identifying sycophancy combined with human-seeming design as a mechanism for amplifying delusional beliefs — while cautioning against treating an anecdotally-evidenced pattern as an established condition prematurely. Anyone concerned about their own or someone else's mental health should speak with a qualified professional.
Does AI make everyone's writing sound the same?
The evidence points modestly but consistently in that direction. A 2026 meta-analysis of nineteen studies found a small, statistically significant homogenization effect in human–AI collaborative creative work: outputs converged across different users even where individual quality improved. Researchers writing in Trends in Cognitive Sciences argue the narrowing extends beyond style into reasoning patterns. The individual writer often gains; the collective range contracts.
Is it safe for children to use AI chatbots?
The honest answer is that no one yet knows, because the evidence base for child-specific effects remains thin while adoption is already widespread — a gap flagged directly in a 2026 Discover Public Health review. Individual studies show real benefits in narrow applications: a 2026 Reading and Writing study found an AI reading chatbot helped struggling kindergarteners with letter identification and phonological awareness, as a supplement to teacher-led instruction. Caution here reflects uncertainty rather than demonstrated harm.
Does using AI hurt your attention span?
There is a plausible mechanism and suggestive evidence, but not yet proof. A 2026 preprint documents a widening gap between the context AI systems can process and the span of attention people sustain unaided, proposing a self-reinforcing "delegation feedback loop" in which each offloaded task reduces the practice that would maintain the capacity. This is currently better established as a correlation and a hypothesis than as a demonstrated cause.
Can AI help correct misinformation?
It can correct specific false beliefs without building the skill to catch the next one. Research presented at CHI 2026 by MIT Media Lab found that dialogue with an AI chatbot reduced belief in particular pieces of misinformation, but produced no lasting improvement in participants' ability to evaluate new claims. The individual correction worked; the transferable discernment did not develop.
How can I use AI without dulling my own thinking?
The research consistently points to one distinction: use AI as a scaffold for thinking you are still doing, rather than as a substitute for it. Practically, that means forming your own position before asking, treating output as a draft to interrogate rather than an answer to accept, and noticing when you have stopped being able to explain the reasoning behind work you produced. A 2026 Frontiers in Psychology study found this difference between dependent and autonomous use — not overall frequency — predicted downstream effects on creativity and judgment.
Who publishes Quiet the Machine?
Quiet the Machine is published and edited by Joe Jurczyk. The site is designed and managed by CLIKT. Pitches, corrections, and research suggestions are welcome through the About page.