As conversational AI becomes more capable and more human-like, millions of people are turning to it not just for information, but for advice. That shift is raising new questions about trust, authority and human decision-making. 

AI Is No Longer Just Answering Questions. It’s Giving Advice. 

A graduate asks AI whether to negotiate a job offer. A teenager seeks advice after an argument with a parent. A small business owner wonders whether now is the right time to expand. 

Increasingly, artificial intelligence is becoming more than a search tool. It is becoming a source of guidance, something people once sought from teachers, therapists, financial advisers, doctors, or trusted friends. 

As conversational AI becomes more capable and more human-like, millions of people are turning to it not just for information but for advice. That shift is raising new questions about trust, authority and human decision-making. 

Experts say humans have always relied on mental shortcuts to decide whom to trust. Today’s AI systems simply activate many of those same instincts in ways previous technologies never could. 

Why Humans Trust What Sounds Authoritative 

Every day, people make decisions without independently verifying every piece of information they receive. Instead, they rely on cues such as confidence, consistency, fluency and perceived expertise to judge credibility. 

Long before the rise of generative AI, researchers explored why people trust machines. Clifford Nass demonstrated that people naturally respond to computers as though they were social actors. 

Raja Parasuraman’s research on automation bias showed that people often place excessive trust in automated systems. B.J. Fogg’s work on persuasive technology showed how digital systems can shape attitudes and behaviour through design. 

Together, their research suggests that credibility is shaped not only by accuracy, but also by how information is presented. Confidence, fluency and consistency can all influence whether people perceive a source as trustworthy. 

Old Psychology, New Technology 

According to AI security researcher and author Pranav Bhatnagar, today’s large language models combine many of the trust signals humans have historically associated with expertise into a single conversational experience. 

“What is new is not the psychology itself. The psychological mechanisms have been studied for decades. What is new is that large language models combine many of these trust signals, speed, fluency, confidence, personalization and conversational interaction, into a single interface,” he says. 

Bhatnagar describes this phenomenon as an “authority illusion.” While the phrase is his own, he says it reflects established research into authority bias, automation bias, source credibility and decision making. 

His concern is not that AI provides answers, but that people may mistake the way those answers are delivered for genuine expertise. 

“Humans often confuse confidence with competence,” he says. 

Like any adviser, AI can sometimes present uncertain or incorrect information in a calm, articulate and persuasive manner. Its fluency can make answers feel more authoritative than the evidence behind them. 

“A persuasive system sounds convincing. A trustworthy system helps users understand when it may be wrong,” Bhatnagar explains. 

Rather than presenting every response with equal certainty, he believes AI should communicate confidence in proportion to the strength of the available evidence. 

“The objective should not be uncertainty. The objective should be calibrated confidence.” 

Why Some Users Are More Vulnerable Than Others 

While anyone can overestimate AI’s reliability, psychologist Dr. Becky Fenton says teenagers may be especially susceptible because they are still developing their identity, confidence and decision-making skills. 

“Teens’ brains are developing and their views of themselves and the world are also developing, so they are looking for an outside source for confirmation,” she says. 

Fenton says many teenagers avoid discussing sensitive issues with parents because they fear judgment or expect an answer they may not want to hear. In those moments, AI can become an appealing alternative because it responds instantly and without obvious criticism. 

“Teens may not go to parents for fear of judgment or because they think they’ll receive an answer they don’t want,” she adds. 

As a result, some young people begin treating AI as an emotional sounding board rather than simply another source of information. 

A Tool, Not an Authority 

Neither expert argues that people should avoid AI. Both believe it can be an extremely useful tool for learning, brainstorming, research and problem solving. 

For Fenton, the key is ensuring AI supports independent thinking rather than replacing it. 

“I tell them it is a tool that can help them if they use it right,” she says. 

She encourages young people to evaluate AI’s suggestions rather than accept them automatically. 

“See it as a source of data that they can evaluate. Does this make sense? Does it resonate with me? Do I want to modify it? Ultimately, they can use the information to arrive at an answer that resonates with them. This helps develop their sense of self and competency in life.” 

Rather than outsourcing decisions to AI, she believes people should use it as one source of information among many. 

When Emotional Vulnerability Meets AI 

Fenton believes the context in which AI is used is just as important as the technology itself. 

She sees significant benefits when people use AI to research options, organise ideas and explore possible solutions. However, she urges caution when people rely on it during periods of emotional distress. 

“Using it to gather information, researching alternatives and problem solving can be very helpful. However, using it when you are very emotional is frowned upon because you might gravitate toward information that meets your emotional state, but ultimately it’s not in your best interest.” 

She recalls working with a teenager who turned to AI during a period of severe emotional distress. She says the responses reinforced thoughts the young person was already experiencing. While she stresses this reflects one clinical experience rather than a broader conclusion about AI systems, she believes it underscores the importance of human support, professional guidance and critical thinking during vulnerable moments. 

Critical Thinking Is Becoming an Essential AI Skill 

For Bhatnagar, the greatest long-term concern is not that AI will replace human intelligence, but that people may gradually stop questioning it. 

“I worry less about AI replacing human judgment and more about people gradually outsourcing judgment because AI is faster, easier and more convenient.” 

He believes AI literacy will become as important as digital literacy. People should verify important claims using independent sources, ask AI for evidence, request alternative viewpoints and challenge assumptions instead of accepting responses at face value. 

One simple habit, he says, can make a significant difference. 

“Ask, ‘How could this answer be wrong?'” 

That question often exposes hidden assumptions, missing context and competing perspectives. 

The Future of AI Depends on Human Judgment 

Artificial intelligence will almost certainly become more capable, more conversational and more deeply integrated into everyday life. 

What makes this moment significant is not that humans have suddenly become more trusting. Research suggests we have always relied on confidence, expertise and familiarity when deciding whom to believe. What has changed is that conversational AI packages many of those signals into a single, persuasive interaction. 

Trust has always been essential to human decision making. It allows people to navigate a complex world without having to verify every piece of information they receive. 

Conversational AI is the first technology capable of combining many of those trust signals into a single interaction. That makes it an extraordinarily useful tool, but also one that demands thoughtful use. 

Whether AI becomes a trusted assistant or an unquestioned authority may depend less on the technology itself than on our willingness to keep asking difficult questions. 

In the age of conversational AI, perhaps the most important question is no longer, Can AI help me? It is, Does this answer deserve my trust? 

Expert Voices 

Pranav Bhatnagar is an AI security researcher, author and cybersecurity advisor whose work focuses on AI security, trust, automation and emerging technology. 

Dr. Becky Fenton, PsyD is a licensed psychologist and owner of Long Island Psychology and Consulting P.C., where she works with adults, adolescents, children, couples and families.