Gartner Says Only 11% Trust AI to Decide: STAR Explains Why

A recent Gartner survey found that consumers want AI to help them shop, but not to make purchase decisions for them. The STAR Framework explains exactly why, and what it means for anyone building, marketing, or deploying AI tools.

Gartner published a finding last month that should make every AI product designer stop and think.

Of the U.S. consumers they surveyed, only 11% are willing to allow AI to make purchase decisions on their behalf. Even for low-risk categories like household supplies and personal care, the number barely moves. Consumers are happy for AI to help them research, compare prices, identify deals, and narrow down their options. But the final call? That stays human.

On the surface, this looks like a trust problem. People don’t trust AI enough to let it decide. And that is partly true: 54% of early adopters reported feeling the need to double-check every piece of information generative AI provides, and 62% considered that information to be a waste of their time. Trust is clearly an issue.

But trust is a surface-level explanation. It tells you what is happening without telling you why. To understand why 89% of consumers want to keep the final decision in their own hands, you need to understand what different types of consumers are actually trying to achieve when they shop, research, or evaluate options. You need to understand their motivational architecture.

That is where the STAR Framework comes in.

The Four Minds Behind the 89%

STAR identifies four core consumer mindsets, each driven by a different fundamental psychological need. The Socialiser is driven by relatedness. The Thinker is driven by competence. The Adventurer is driven by autonomy. The Realist is driven by security. These are not personality quirks. They are deep motivational structures that shape how people process information, evaluate risk, and make decisions.

When you map the Gartner findings onto these four types, the 89% stops being a monolithic block of “AI sceptics” and becomes four distinct groups, each resisting AI decision-making for completely different reasons.

The Thinker: “I Will Verify It Myself”

The Thinker is the most visible type in the Gartner data, and the most immediately relevant to the AI industry’s challenges.

The finding that 54% of early adopters feel the need to double-check every AI output is textbook Thinker behaviour. The Thinker’s core psychological need is competence, which they satisfy through the act of rigorous, independent analysis. They do not just want the right answer. They want to have arrived at the right answer through their own process of evaluation. The rigour is the point.

When a Thinker uses an AI tool to research a product, they are not delegating the analysis. They are gathering inputs for their own analysis. The AI is one data source among many, and it carries no inherent authority. The Thinker will cross-reference it against reviews, expert opinions, their own experience, and probably two or three other AI tools. If the AI’s recommendation does not survive this scrutiny, it is discarded. If it does survive, the Thinker takes ownership of the conclusion, not the AI.

The Gartner finding that 62% of early adopters consider AI information “a waste of their time” is not anti-AI sentiment. It is Thinkers discovering that the tool does not actually save them work. If the AI produces a recommendation that still needs to be fully verified, the Thinker has not saved time. They have added a step. The tool is creating more work than it eliminates, and for a type whose entire decision-making identity is built on rigour, that is a product failure, not a user failure.

The strategic implication for AI product design is clear: tools built for Thinkers must be transparent about their reasoning. Show the sources. Show the methodology. Show the confidence intervals. A Thinker who can see how the AI reached its conclusion can evaluate the reasoning without having to redo the entire analysis from scratch. A black-box recommendation is useless to a Thinker, not because they are technophobic, but because a recommendation without reasoning is not competence. It is faith.

The Realist: “Help Me Choose, Don’t Choose for Me”

The Realist is the pragmatist in the Gartner data. They are the 31% willing to let AI narrow choices for household supplies, and the 28% willing to let it narrow choices for personal electronics. Their relationship with AI is transactional and conditional.

The Realist’s core psychological need is security. They want reliable outcomes, minimal risk, and predictable processes. AI fits into this framework as a research assistant, not a decision-maker. The Realist is perfectly happy for AI to surface the best options, compare prices, and flag deals. What they are not happy about is the AI making the final call, because the final call is where the risk lives.

For a Realist, the decision to purchase is not just about the product. It is about the process. They want to know that they have considered the options, that they have not missed something important, and that the outcome will be what they expect. Delegating the decision to an AI removes their ability to verify these things. It introduces uncertainty into a process that the Realist needs to feel certain about.

The Gartner finding that consumers want AI to help them “find better information, compare prices, and identify deals” is almost a perfect description of the Realist’s ideal AI tool. It is a tool that reduces the workload of due diligence without removing the human’s role in the final evaluation. The Realist does not want the AI to think for them. They want the AI to give them better things to think about.

The product design implication: Realists need AI tools that are honest about what they know and what they do not know. A recommendation engine that says “here are three options that match your criteria, ranked by price and customer rating” is useful to a Realist. A recommendation engine that says “buy this one” is not. The difference is not the information. The difference is whether the tool respects the Realist’s need to make the call themselves.

The Socialiser: “What Do People I Trust Think?”

The Socialiser is the type most likely to be in the 72% of consumers who report passive exposure to generative AI through their internet and app usage. They encounter AI through social media algorithms, recommendation engines, collaborative tools, and shared shopping experiences. AI is woven into the fabric of how they connect with other people.

The Socialiser’s core psychological need is relatedness. They evaluate products, services, and experiences through the lens of how those things affect their relationships and social connections. When a Socialiser shops, they are not just buying a product. They are participating in a social process. They want to know what their friends think, what people they trust recommend, and what will bring them closer to the people they care about.

This is why the Socialiser is the type most likely to delegate some of the decision to AI, but only if the AI is connected to social proof. A recommendation engine that says “people like you also bought this” is appealing to a Socialiser, not because they trust the algorithm, but because the algorithm is proxying the opinions of people like them. The AI is not making the decision. The community is making the decision, and the AI is just the messenger.

The Gartner finding that only 11% want AI to make decisions, even among the most socially-oriented consumers, tells us something important: even for Socialisers, the social process of decision-making is part of the value. Buying something because an algorithm said to is not the same as buying something because your friend recommended it. The Socialiser wants the connection, not just the outcome.

For AI product designers, this means building social layers into recommendation tools. Show what people in the user’s network have purchased. Show reviews from people with similar profiles. Make the recommendation feel like it comes from a community, not a machine. The Socialiser will accept AI-mediated recommendations if those recommendations feel socially validated.

The Adventurer: “Don’t Pre-Filter My World”

The Adventurer is the type most likely to resist AI-mediated shopping entirely, and the Gartner data explains why.

The Adventurer’s core psychological need is autonomy. They want to discover things for themselves, forge their own path, and make decisions that feel authentically their own. The act of shopping, for an Adventurer, is not a problem to be optimised. It is an experience to be had. They want to browse, stumble upon things they did not know existed, and make serendipitous discoveries. An AI that pre-filters their options, even with the best intentions, is removing the very experience they value.

The Gartner finding that consumers want AI to “narrow selections” is, for an Adventurer, a description of the problem, not the solution. Narrowing selections is what the Adventurer does not want. They want the full range of options, including the unexpected ones. They want to be surprised. They want to find something they were not looking for. An AI that learns their preferences and serves them a curated feed is, from the Adventurer’s perspective, a cage dressed up as a convenience.

This is why the 11% figure is so low. Even among consumers who are generally comfortable with technology, the Adventurer’s need for autonomy creates a hard boundary. They will use AI for research, because research is a means to an end. They will not use AI for discovery, because discovery is the end itself.

For AI product designers, the Adventurer is the hardest type to serve. The instinct is to personalise, to learn preferences, to narrow and curate. But every act of curation is an act of constraint from the Adventurer’s perspective. The best AI tools for Adventurers are the ones that expand options rather than narrow them. “Here are ten things you might not have considered” is more useful to an Adventurer than “here is the one thing you should buy.”

The 62% Problem: When AI Fails to Match Motivation

The most STAR-relevant number in the Gartner data is not the 11%. It is the 62% of early adopters who consider AI-generated information to be a waste of their time.

This is not a technology problem. It is a motivation problem. The AI is producing information, but it is not producing the type of information that different consumer types need to feel satisfied with the process.

For the Thinker, the information lacks transparency. They cannot see the reasoning, so they cannot verify it, so it creates more work than it saves.

For the Realist, the information lacks reliability signals. They cannot tell whether the AI’s recommendation is trustworthy, so it does not reduce their risk. It increases it.

For the Socialiser, the information lacks social context. A recommendation from a machine is not the same as a recommendation from a person they trust. Without the social layer, the information feels sterile and unconvincing.

For the Adventurer, the information is too narrow. It reflects what they already like, not what they might discover. It confirms their preferences rather than challenging them.

In every case, the AI is failing not because its output is technically wrong, but because its output does not match the motivational architecture of the person receiving it. The same recommendation, delivered in the same way, produces four completely different reactions depending on the type of consumer receiving it.

What This Means for AI Strategy

The Gartner survey is a wake-up call, but not in the way most people will interpret it.

The standard reading is: “Consumers don’t trust AI enough. We need to build more trust.” That is true, but it is incomplete. Trust is not a single variable. It is a composite of different needs, different expectations, and different decision-making styles. Building trust with a Thinker requires transparency. Building trust with a Realist requires reliability. Building trust with a Socialiser requires social validation. Building trust with an Adventurer requires respecting their autonomy.

The STAR reading is: “Consumers don’t trust AI because AI is being deployed as a one-size-fits-all solution for a population with four fundamentally different motivational architectures.”

The 11% who are willing to let AI decide are not a target market to be expanded. They are a signal. They represent the small minority whose motivational needs happen to be served by the current generation of AI tools: likely Socialisers who experience AI as a proxy for community recommendations, or Realists who have done the risk calculus and decided the efficiency gain is worth it. Growing beyond 11% requires not more AI, but better-matched AI.

The Gartner recommendation to focus AI investments on tools that “facilitate consumer research, price comparisons, deal discovery, and choice narrowing” is sound advice. But it is advice without a framework. STAR provides the framework. It tells you why different consumers research differently, what each type needs from the tool to trust it, and how to design AI experiences that feel helpful rather than intrusive.

The future of AI in consumer decision-making is not autonomous agents making purchases on behalf of humans. It is intelligent tools that understand the human they are serving well enough to match their support to the person’s motivational architecture. The Thinker gets transparency. The Realist gets reliability. The Socialiser gets social proof. The Adventurer gets expanded options.

Same technology. Four different experiences. That is the STAR difference.

David Chadderton is the creator of the STAR Framework and the author of The STAR Framework: Rewriting the Rules of Consumer Engagement (NYC Big Book Award 2025), The STAR Operating System, and Dear Algorithm, It’s Not Me, It’s You. He writes about human behaviour, AI, and the psychology of decision-making on The Unoptimised Human.