Revealing The Many Dimensions Of Psychologically Trusting AI

The Minds Problem

Much of the handwringing about AI is whether AI is going to become sentient and embody consciousness, often referred to as AI as a thinking machine. That is an important topic and one that I’ve continued to closely explore and analyze; see the link here. There is a quite different angle that also deserves rapt attention, and I’d like to place it on the table.

How will human minds change because of interacting with AI?

This unexpected question catches many people by surprise. The customary focus is on AI as formulating a mind, not on how human minds might change due to interacting with AI. But, as the old saying goes, it takes two to tango. When humans increasingly interact with AI, there is a two-way street involved. Humans adjust their minds about how they view the world and how to communicate with others, and shift in ways that we still do not have a full or clear picture of.

In that sense, it is wholly worthwhile to study the psychology of humans as they mentally adjust to a world that entails human-to-AI interaction and human-to-human interaction. Notably, this doesn’t have to wait until or if AI becomes sentient. There is plenty to study right now. Humans are already adjusting their minds to the ubiquitous nature of non-sentient AI. The starting gun has gone off. Human minds are changing.

For my comprehensive tracing of how AI and psychology dovetail with each other, see the link here and the link here.

Human Trust In AI

Shifting gears, let’s discuss the human-AI trust bond. There are an estimated 1.5 billion people weekly using generative AI and large language models (LLMs) for all sorts of daily tasks, ranging from the mundane to highly sensitive needs (see my analysis of global AI usage at the link here).

Do people trust what AI is doing? That’s an important question to ask since a notable portion of the population seems to be significantly relying on AI.

There is an interesting psychological relationship between having trust in AI and a secondary factor consisting of the amount of trust that goes towards the AI maker who developed the AI. In other words, a person’s trust in AI is seemingly moderated by whether they trust the AI maker. If an AI maker is not considered trustworthy, you would naturally tend to be less trusting of the AI that they have devised. Likewise, a high trust in an AI maker would tend to lean you toward being trusting of the AI they’ve made. I’ve previously posted an in-depth exploration of institutional AI maker trust and capability AI trust at the link here.

Trust Arises In Many Dimensions

It is fair to say that there are at least two dimensions associated with trusting AI, consisting of institutional trust and capability trust. These are two crucial dimensions. But we need to widen our scope and consider additional dimensions that rightfully deserve close attention too.

Often, psychological research on trusting AI tends to focus on one or maybe two dimensions. There is no doubt that this is useful research. Kudos for tackling an evolving and vexing topic. At the same time, this often inadvertently misses a bigger-picture perspective. The implication of studies that focus on one or two dimensions is that those are the only dimensions worthy of examination. Nope, there are many more that deserve their moment in the sunshine too.

I will walk you through a multitude of dimensions that permeate the AI trust conundrum. The collective set is a veritable portfolio of dimensions. They all matter. I’m not suggesting that all psychological research on trust in AI must necessarily address all of them in one fell swoop. It is perfectly sensible to concentrate on a subset of dimensions. The key is to dutifully articulate that additional dimensions exist and might be integral or at least notable regarding the selected dimensions being assessed.

Crucial List Of Human-AI Trust Dimensions

The mainstay dimensions underlying human-AI trust consist of these seven:

• Dimension #1: Institutional trust. Do you trust the AI maker and/or AI deployer?
• Dimension #2: Capabilities trust. Do you trust that the AI can do the tasks or answer the questions you pose to it?
• Dimension #3: Accuracy trust. Do you trust that the AI is giving you reliable answers?
• Dimension #4. Integrity trust. Do you trust that the AI is being truthful rather than deceptive?
• Dimension #5: Benevolence trust. Do you trust that the AI is trying to be helpful to you versus seeking to harm you?
• Dimension #6: Privacy trust. Do you trust that the AI will keep your interactions private?
• Dimension #7: Governance trust: Do you trust that the AI is being properly governed to protect you and perform safely?

Allow me a moment to showcase how picking just two dimensions can fail to tell the whole story. Suppose that a person named Jane is considering using a popular AI, such as ChatGPT, GPT-5, Grok, Claude, Gemini, Copilot, etc. Jane has heard that the AI maker has an excellent reputation, so this ratchets up the institutional trust dimension. Friends of Jane have said that the AI is very good at answering questions about mental health. Since Jane is primarily interested in getting mental health guidance from AI, the rave reviews by friends have raised trust in the capabilities dimension.

So far, so good. Jane perceives the AI as being high on institutional trust and high on capabilities trust. Time to go ahead and dip into the AI with some mental health questions.

Scenario Involving More Dimensions

Jane starts using the AI. Numerous in-depth personal dialogues take place on mental health topics. The trust that Jane placed in the AI seems completely warranted. Trust remains high. Nice.

Suddenly, Jane becomes concerned during a dialogue. Jane’s trust in the AI drops like a rock. Was it due to a reduction in institutional trust, such as maybe hearing on the news that the AI maker has gotten into financial trouble? No, that didn’t weigh into the trust lessening. Was it because the AI itself seemed unable to answer Jane’s questions about mental health? No, that wasn’t an issue and didn’t have an impact on capabilities trust.

What was the culprit that knocked the perceived trust to the ground?

Turns out that the AI indicated that it was automatically going to alert an external safety team about Jane’s interactions. An area of mental health discussion was about the nuances of self-harm. This triggered an aspect in the AI that necessitates letting human agents know that Jane has been engaging in a dialogue of that nature. The idea is that this is a precautionary aspect that the AI maker has implemented after lawsuits against the company contended that the AI lacked timely reporting on self-harm indications.

Dimensions Touched Upon

You can likely see that the institutional trust might remain high since the AI maker appears to be doing something on behalf of users on a safety basis. The level of capabilities trust remains high due to the AI still answering Jane’s questions on mental health topics, including the sensitive topic of self-harm.

But from Jane’s perspective, this sending of an alert by the AI to human outsiders is a breach of privacy trust. Thus, despite retaining institutional trust and capabilities trust, the break in privacy trust is going to outweigh those trusts in this instance. Jane did not realize that the conversations with the AI would be shared with anyone else. Jane had assumed that the AI would forever keep any human-AI interactions utterly private.

Governance trust takes a bit of a hit too. Why? Though the AI maker does seem to be giving proper credence to governance of the AI, in Jane’s viewpoint, the AI has made a snap judgment. Jane has no intention of pursuing self-harm. It was just a topic of interest. Plus, Jane had a family member who had previously suffered from self-harm. The AI opted to turn this chat into an alarm when it was nothing of the kind.

In this scenario, institutional trust and capabilities trust remained high. If only those two dimensions were being used to explain why Jane has radically reduced trust in the AI, we would be stumped. By realizing that there are additional dimensions underlying trust in AI, we could examine those other dimensions. In this circumstance, privacy trust took a huge hit, and governance trust took a sharp hit. The overall level of trust fell. The height of the institutional trust and capability trust was not enough to shore things up.

Portfolio Of Trust Dimensions

Trust in AI is not a single variable or unidimensional. If you try to pin the psychological formulation of trust on one or two dimensions, the odds are that those dimensions will not adequately cover all the bases. Circumstances will arise in which those one or two dimensions do not sufficiently capture why trust has risen or fallen.

In fact, it might be confusing and head-scratching since the one or two dimensions might remain high, yet the trust of the person has dropped precipitously. The same can occur in the other direction. Trust might be low in those dimensions, and nonetheless the AI may be considered to be high trust in the mind of the person.

This is an important consideration for those researching why people trust or distrust AI. Make sure to encompass the full suite of dimensions. AI makers also would be wise to give prudent attention to all the trust dimensions. If an AI maker concentrates on only one or two and applauds themselves for how they are managing trust in their AI, they are sadly going to have a rude awakening. A dimension of trust that they didn’t know existed and weren’t maintaining could sink their whole ship.

Trust In AI Is A Big Deal

People won’t use AI if they believe that AI is not trustworthy. A large enough lack of trust could wipe out an AI maker. Widespread distrust could wipe out the entire AI industry. Though AI makers are flying high right now, it seems entirely feasible that trust could abruptly evaporate, and society decides that AI can no longer be trusted. Policymakers and lawmakers would almost certainly put in place draconian laws that would put a stop to AI.

Trust in AI is not something to be idly treated or sloppily taken as a given.

A final thought for now. The famous painter and ceramicist Walter Inglis Anderson famously made this remark: “Trust is like a vase. Once it’s broken, though you can fix it, the vase will never be the same again.” AI makers need to realize that they won’t easily overcome a major bout of distrust. Right now, they are still in a societal honeymoon stage. A trust fall could be around the next bend. Be watchful, be mindful.