Americans use AI more often, but trust the results less and less — Quinnipiac poll

The more Americans begin using AI tools, the less they trust them. This is a paradoxical but logical conclusion from a new Quinnipiac poll: technology adoption is growing steadily, but confidence in the accuracy and reliability of its results is not. Quinnipiac University is one of the leading American polling centers, regularly conducting large-scale surveys on current public issues. This time, analysts investigated how ordinary Americans perceive artificial intelligence: whether they use it at work and in their personal lives, whether they trust the answers they receive, and whether they want to see government regulation in this sphere. The answers turned out to be contradictory.

The results reveal a pattern characteristic of many technological revolutions: adoption outpaces understanding. People begin applying AI tools—for writing text, searching for information, analyzing data, communication, and making everyday decisions—before they have time to form a clear understanding of how much these systems can actually be trusted. The outcome is predictable: the more experience with use, the sharper the sense of uncertainty and the more often questions arise.

Three main themes of concern identified by the poll are transparency, regulation, and AI's broad impact on society.

Transparency is a long-standing pain point for the entire industry. Most modern AI systems work like black boxes: the user sees the result but does not understand how exactly the model reached it, what data it was trained on, and what distortions may be built into it. This is particularly critical in medicine, law, and finance—fields where an AI error can directly cost a person their health or money. When the decision-making mechanism is opaque, trust in it remains fragile by definition.

Regulation is a question that has been discussed at all levels for several years now, from state legislatures to international forums. So far, there is no unified federal framework in the US: individual states are introducing their own rules, the European Union has adopted the AI Act, but an American AI law as such does not yet exist. The absence of clear rules of the game strengthens distrust: people do not know who is responsible if something goes wrong and where to file a complaint.

Broad impact on society is the most abstract but emotionally charged topic. Job loss, the spread of deepfakes and disinformation, the concentration of enormous power in the hands of a few tech corporations, influence on elections and the media landscape—all of this falls within the scope of public concern. People do not simply distrust the specific answers of a specific chatbot. They doubt the direction in which technology as a whole is moving and who actually controls it.

The gap between adoption and trust is a non-trivial situation. Telephones, the internet, smartphones—all of these technologies passed from mass skepticism to everyday norm, gradually gaining trust through familiarity. With AI, this scenario may not work or may work significantly slower. The reason lies in the nature of the technology itself: AI imitates human thinking. Trusting a tool that thinks for you is psychologically much harder than trusting a tool that simply accelerates a familiar process.

For companies and developers, the poll results are a direct signal: technological progress without work on trust becomes unrealized potential. Algorithm explainability, independent audits, honest discussion of limitations, and clear accountability mechanisms—this is what the industry needs right now just as much as the next updated version of a model.