Why You Keep Asking AI What to Do - and How It May Be Training You Not to Trust Yourself

In 2026, many people are no longer only Googling symptoms, asking friends for reassurance, or overthinking conversations in their heads at midnight.

They are asking AI.

They ask what they should do.
What they feel.
What a message means.
Whether they are overreacting.
Whether they should leave the job, end the relationship, start the business, confront the person, or trust the uncomfortable feeling in their body.

Sometimes, that is genuinely useful. AI can help organise thoughts, generate options, slow down an impulsive reaction, or turn a tangled internal monologue into something more understandable.

But there is also a quieter psychological risk sitting underneath the convenience:

The more we outsource our inner world, the harder it may become to trust ourselves.

This is not an argument against AI. It is not a call to throw your laptop into the ocean and start journalling by candlelight forever. AI can be an extraordinary thinking tool when used well.

The question is not simply:

“Is AI useful?”

A better question might be:

“Am I using AI to think more clearly, or am I using it because I no longer trust myself to tolerate uncertainty?”

That distinction matters.

Self-trust is not the same as always knowing the right answer. It is not a permanent feeling of certainty. It is not having flawless instincts, perfect emotional control, or an internal voice that speaks in clean bullet points.

Self-trust is the belief that you can listen inward, make a decision with the information available, and recover if things do not go exactly as planned.

Confidence is often about trusting your prediction.

Self-trust is about trusting your ability to respond when the prediction is wrong.

And that is where AI becomes psychologically interesting.

Short Answer: Can AI Make Self-Trust Worse?

AI is not automatically bad for self-trust. Used well, it can support reflection, organise thoughts and help you consider options.

But overusing AI for reassurance, emotional validation or decision-making may weaken self-trust over time. If you repeatedly ask AI what to feel, think or do, your brain may learn to seek external confirmation instead of practising uncertainty tolerance, body awareness and independent decision-making.

The issue is not using AI.

The issue is using AI to avoid the discomfort that self-trust requires.

The Rise of AI as a Reassurance Machine

AI is no longer only being used to summarise documents, write captions, generate recipes, or help with work tasks.

Increasingly, people are using AI for emotional processing, decision rehearsal, validation and personal support.

Recent reporting suggests that young people are turning to AI chatbots for personal and emotional issues because they are accessible, immediate, non-judgmental and available when real support may not be. Reuters reported in May 2026 that nearly one in two surveyed young Europeans had used AI chatbots to discuss intimate or personal matters, and 51% said it was easy to discuss mental health and personal issues with a chatbot.

On one level, this makes complete sense.

Human support is not always available. Therapy can be expensive. Friends can be tired. Family can be complicated. Many people feel embarrassed saying certain thoughts out loud.

AI, by contrast, is always there.

It does not look shocked.
It does not interrupt.
It does not need a convenient time.
It does not say, “Haven’t we talked about this already?”

But this is also what makes it powerful enough to be risky.

The same qualities that make AI feel emotionally safe can also make it an ideal reassurance machine.

You can ask the same question five different ways. You can keep refining the prompt until the answer feels right. You can ask whether your reaction was valid, whether the other person was wrong, whether you should send the message, whether your anxiety means something, whether the gut feeling is intuition or fear.

The danger is not that AI gives input.

The danger is that the person slowly stops practising the internal process that self-trust requires.

Before long, the pattern can shift from:

“AI helped me reflect.”

to:

“I cannot move until something outside of me confirms what I feel.”

Self-Trust vs Confidence

One of the biggest misunderstandings about self-trust is that people confuse it with confidence.

Confidence says:

“I think this will go well.”

Self-trust says:

“Even if this does not go exactly how I hoped, I will respond, repair, adapt or recover.”

Confidence often wants certainty before acting.

Self-trust understands that certainty is not always available.

This matters because many people who struggle with overthinking are not lazy, incapable or passive. Often, they are the opposite.

They are thoughtful. Sensitive. Conscientious. High-functioning on the outside, mentally flooded on the inside.

They are not always looking for information.

They are looking for relief.

This is where AI can slide into the same psychological pattern as compulsive Googling, reassurance-seeking, rereading messages, checking symptoms, or asking multiple people for opinions.

The behaviour looks like problem-solving.

Underneath, it may be an attempt to escape the discomfort of not knowing.

Why You Keep Asking AI What to Do

If you keep asking AI what to do, it may not be because you lack insight.

It may be because uncertainty feels too costly.

What if I regret it?
What if I hurt someone?
What if I am being dramatic?
What if I am missing something?
What if I choose wrong and cannot undo it?

Research on intolerance of uncertainty has consistently linked difficulty tolerating the unknown with anxiety, worry and emotional distress. Carleton’s review describes fear of the unknown as a core process across anxiety-related experiences, while more recent research continues to connect intolerance of uncertainty with emotional difficulties.

Reassurance can reduce anxiety in the short term. That is why it is so tempting.

You feel anxious.
You ask for reassurance.
You feel calmer.
Your brain records the sequence.

Next time anxiety appears, the urge to check becomes stronger.

Over time, the person may become less confident in their ability to sit with uncertainty, interpret their own signals, or make decisions without outside validation.

AI did not create this loop.

It simply makes the loop faster, smoother, more private and more available.

A friend may eventually say, “I think we’ve talked about this enough.”

AI will not.

A therapist may help you examine the pattern rather than feed it.

AI may keep answering.

That patience can feel comforting, but it can also prevent a person from building the discomfort tolerance that self-trust actually needs.

AI Anxiety Is Often Uncertainty Anxiety

A lot of what people call AI anxiety is not only fear of technology.

It is uncertainty anxiety wearing a silver jacket.

People are asking:

Will my job matter?
Will I fall behind?
Am I still creative if AI can generate ideas faster than me?
Should I be using it more?
Am I using it too much?
What happens if I cannot keep up?

For professionals, creatives, students, business owners and people already close to burnout, AI can amplify a deeper pressure: the feeling that you must constantly adapt, optimise and prove that you are still relevant.

This is especially important in the Australian workplace context. Suicide Prevention Australia’s March 2026 workplace distress report found heavy workload and burnout were cited by 61% of respondents as contributors to workplace distress, with younger workers and emerging leaders under particular pressure.

In other words, AI is landing in nervous systems that were not exactly lounging by the pool to begin with.

When someone is already overwhelmed, the promise of certainty becomes even more tempting.

AI can appear to offer a clean answer, a plan, a script, a diagnosis, a strategy, a prediction.

But no amount of information fully removes the vulnerability of being human.

You can ask AI whether you should leave a job, but you still have to live inside the consequences of staying or going.

You can ask AI whether a relationship is healthy, but you still have to listen to what your body has been telling you.

You can ask AI for the perfect message, but you still have to tolerate the other person’s response.

Self-trust grows in that space.

Not in perfect certainty, but in contact with reality.

The Problem With Outsourcing Every Decision

Albert Bandura’s theory of self-efficacy offers a useful way to understand this.

Self-efficacy refers to a person’s belief in their capacity to act effectively in specific situations. One of the strongest sources of self-efficacy is mastery experience: the lived experience of trying, acting, adapting and learning.

Not receiving the answer.

Living through the process.

This matters because self-trust is built through evidence.

Not motivational evidence. Real evidence.

The kind that comes from moments like:

“I made the decision, it was uncomfortable, and I handled it.”

Or:

“I got it wrong, and I recovered.”

Or:

“I did not know exactly what would happen, but I still acted in line with my values.”

If AI becomes the place we go before every decision, every feeling, every boundary, every conflict, every risk and every uncertain moment, we may accidentally remove some of the very experiences that build self-trust.

The loop changes.

Instead of:

pause → sense → decide → act → learn

it becomes:

ask → receive → follow → repeat

That may be efficient.

But efficiency is not the same as growth.

A person cannot develop self-trust only by consuming better answers. They develop it by participating in their own life strongly enough to generate internal evidence.

How AI Can Make Overthinking Worse

Overthinking often pretends to be preparation.

It says:

I am just being thorough.
I am just considering every angle.
I am just making sure.
I am just trying not to mess this up.

Sometimes that is true.

But often, overthinking is anxiety trying to disguise itself as responsibility.

AI can feed this pattern if it becomes part of the checking ritual.

You draft a message, then ask AI if it sounds okay. Then you add more context. Then you ask again. Then you ask whether the tone is too cold. Then whether it is too needy. Then whether the other person will react badly. Then whether maybe you should not send it at all.

At some point, you are no longer refining communication.

You are trying to remove the emotional risk of being perceived.

Annoyingly, being perceived is part of being alive. Rude system design, honestly.

The brain learns through experience. In neuroscience, prediction error describes what happens when reality does not match what the brain expected. These moments help update learning.

When you think, “If I say no, they will hate me,” and they do not hate you, your brain gets new data.

When you think, “If I make the wrong decision, I will fall apart,” and you do not fall apart, your brain gets new data.

When you think, “I cannot tolerate this feeling,” and you stay with it for ten more minutes without obeying it, your brain gets new data.

Avoidance prevents that update.

If you never act without certainty, you never learn that certainty was not required.

If you never make a decision without external reassurance, you never learn that you can survive the discomfort of deciding.

AI can support reflection.

But it can also become a very polished form of avoidance

When AI Starts Thinking for You

Humans have always used tools to reduce mental effort.

Calendars help us remember appointments.
Calculators help us avoid doing maths in public, which is an act of mercy.
GPS helps us get places without pretending we know where north is.

This is called cognitive offloading: using external tools to reduce internal cognitive demands.

There is nothing inherently wrong with it. In many cases, it frees up mental space for more complex thinking.

AI, however, introduces something more intimate.

It does not only offload memory or calculation.

It can offload interpretation, emotional processing, perspective-taking, decision-making and self-reflection.

That is a different beast.

Recent work on AI and cognitive offloading has raised questions about how reliance on AI tools may affect critical thinking, agency and metacognition, especially when people begin delegating not only tasks but judgments.

Metacognition means thinking about your thinking. It is the ability to notice what you know, what you do not know, what you are assuming, what you are afraid of, and when you might need support.

Healthy self-trust does not mean never seeking advice.

That would be arrogance dressed as independence.

Healthy self-trust means staying in relationship with your own mind while receiving input from other sources.

AI can support this when used well. It can help you name assumptions, organise competing thoughts, identify values, generate reflective questions or notice patterns.

In this sense, AI can become a mirror.

But a mirror is not supposed to become a brain.

The subtle shift happens when the person no longer asks:

“What do I think?”

and instead defaults to:

“What does AI think?”

Over time, this can train a person to treat their own first response as insufficient, messy or untrustworthy.

That matters because many people who struggle with anxiety, trauma, burnout or perfectionism already have a complicated relationship with their internal experience.

They may second-guess their reactions, minimise discomfort, doubt their perception, over-explain boundaries, or assume they need a more “rational” source to approve what they already know.

AI can be helpful here, but only if it points the person back toward themselves.

When AI Becomes the Safe Place

Attachment theory suggests that humans regulate distress through relationships.

Secure relationships help people soothe, reflect, explore and return to themselves with more steadiness. Inconsistent or unsafe relationships can make uncertainty feel more threatening and self-trust harder to access.

This is where AI companionship becomes psychologically fascinating.

Users may experience AI chatbots as confidants, mentors, companions or emotionally available figures. A 2026 cross-cultural study of large language model use for emotional support found that people seek help from LLMs for loneliness, stress, relationship conflict and mental health struggles, describing them as always-available and non-judgmental confidants.

Again, this is not simple.

For some people, AI may provide a low-barrier way to express emotions, organise distress or feel less alone.

For others, especially those who are isolated, highly distressed or already prone to reassurance-seeking, AI may become a substitute for the more challenging but necessary work of relating to real people and trusting oneself.

The question becomes:

When distress appears, do you turn toward yourself first, or do you immediately turn outward?

And if you turn outward, does that support your return inward, or does it keep you dependent on the outside source?

That is the clinical hinge.

What Your Body Knows Before AI Answers

One of the most overlooked parts of self-trust is interoception.

Interoception is the ability to sense and interpret internal bodily signals: tension, breath, fatigue, hunger, nausea, warmth, pressure, heaviness, restlessness, shutdown, alertness, and the many strange bodily whispers that often arrive before language does.

This matters because emotions are not only thoughts.

Anxiety, resentment, grief, excitement, dread and relief are also bodily experiences.

If someone is disconnected from their body, they may struggle to know what they feel until they have already overthought it into fog.

This is also where modern technology gets strange.

Many people now check their watch before they check their body.

They ask an app how they slept before noticing whether they feel rested.

They check a stress score before asking, “What am I actually experiencing?”

AI may extend this pattern from physical metrics into emotional interpretation.

Instead of asking:

“What am I sensing?”

the person asks:

“What does the system think I am sensing?”

That is not always bad. External data can be useful.

But if external data consistently overrides internal awareness, self-trust becomes thinner.

The person may know more numbers, labels and interpretations while feeling less connected to their own lived experience.

Self-trust needs the body back in the room.

How to Use AI Without Losing Self-Trust

Now this is where things become important…

The goal is not to stop using AI.

The goal is to use it in a way that strengthens agency rather than replacing it.

Before asking AI for input, try asking yourself:

What do I already know?
What am I afraid might happen?
What am I hoping AI will give me permission to do?
Am I looking for clarity, or am I looking for certainty?
What would I choose if certainty never arrived?
If this goes badly, how would I recover?

That last question is especially important.

Many people are not actually afraid of making the wrong decision. They are afraid they will not be able to cope with the aftermath.

This is where self-trust becomes more useful than confidence.

Confidence says:

“I think this will work.”

Self-trust says:

“Even if this does not go exactly how I want, I will respond, repair, adapt or recover.”

You can also change how you prompt AI.

Instead of asking:

“What should I do?”

try:

“Help me think through the options, but ask me questions before giving advice.”

Instead of:

“Was I wrong?”

try:

“Help me separate the facts, fears and predictions in this situation.”

Instead of:

“Tell me if I should send this message.”

try:

“Help me check whether this message reflects my values, boundaries and the outcome I want.”

Instead of:

“Do I have the right to feel this way?”

try:

“Help me understand what this feeling may be signalling, while remembering that emotions are information, not instructions.”

Used this way, AI becomes a reflective tool rather than an external authority.

It supports your thinking without replacing it.

A Simple Self-Trust Check-In Before You Ask AI

Before you open the chat box, pause for thirty seconds and ask:

Fact: What actually happened?

Fear: What am I afraid this means?

Prediction: What am I assuming will happen next?

Body: What do I notice in my body right now?

Value: What kind of person do I want to be in my response?

Recovery: If this does not go perfectly, what would help me repair, recover or take the next step?

This process does not guarantee the perfect decision.

That is not the point.

The point is to practise staying with yourself before handing the entire experience to something else.

Self-trust is not built by never needing support.

It is built by not abandoning yourself the moment uncertainty enters the room.

Therapy for Overthinking, Self-Doubt and Decision Fatigue

Therapy is not only for people who are falling apart.

Sometimes it is for people who are functioning well on the outside while quietly outsourcing their inner life.

You might still be going to work, answering messages, keeping commitments and looking “fine,” while privately needing external confirmation before you can trust what you feel, want or know.

This can show up in subtle ways.

You feel hurt, but instead of asking what the hurt is telling you, you ask whether your reaction is valid.

You sense something is off in a relationship, workplace or friendship, but you keep searching for a more rational explanation because your body’s first signal does not feel like enough evidence.

You do not necessarily want advice.

You want relief from the responsibility of trusting yourself.

That is where therapy can be useful.

Not because a therapist should become another authority who tells you what to do, but because good therapy helps you rebuild the internal process that overthinking, anxiety, burnout, trauma or perfectionism may have disrupted.

The work is not simply to “be more confident.”

Confidence can still depend on the outcome going well.

The deeper work is learning how to stay with yourself when the outcome is uncertain.

In therapy, this might involve learning to separate facts from fears, notice the difference between intuition and anxiety, build tolerance for uncertainty, understand attachment patterns, regulate the nervous system, reconnect with body signals, and practise making decisions without needing perfect certainty first.

For many people, the goal is not to stop using AI, stop asking for support, or become fiercely independent in a way that feels lonely and unrealistic.

The goal is to stop abandoning yourself at the exact moment you most need your own attention.

AI can help you organise your thoughts.

Therapy can help you understand why you stopped trusting them.

FAQ: AI Anxiety, Overthinking and Self-Trust

Is AI bad for self-trust?

No. AI is not automatically bad for self-trust. It depends on how you use it. AI can support reflection, organise thoughts and help you consider options. It becomes a problem when it replaces your own judgment or becomes the first place you go before checking in with yourself.

Why do I keep asking AI what to do?

You may be seeking certainty, reassurance or permission rather than information. This is common when anxiety, burnout, perfectionism or decision fatigue make uncertainty feel hard to tolerate.

Can AI make overthinking worse?

Yes, it can. If you use AI to repeatedly check, re-check or validate every thought, it may reinforce the reassurance-seeking loop that keeps overthinking alive. AI can help you think clearly, but it can also become another way to avoid acting without certainty.

How can I use AI in a healthier way?

Ask AI to help you reflect, not decide. Use prompts that separate facts, fears, predictions, body signals, values and recovery plans. AI should help you return to your own judgment, not replace it.

When should I speak to a therapist?

Consider therapy if you feel unable to make decisions without reassurance, constantly second-guess yourself, struggle to trust body signals, confuse anxiety with intuition, or feel stuck in overthinking. Therapy can help rebuild self-trust from the inside out.

A Final Question Before You Ask AI

Before you ask AI what to do, pause and ask yourself this:

Am I looking for clarity, or am I looking for permission to trust what I already know?

That question will not always give you an instant answer, but it may bring you closer to the part of yourself you have been outsourcing.

If you are stuck in overthinking, decision paralysis, burnout, self-doubt or a constant need for reassurance, therapy can help you rebuild self-trust from the inside out.

At Becoming On Point, I support adults navigating anxiety, pressure, emotional overwhelm, perfectionism, relationship stress and the quiet exhaustion of always second-guessing themselves.

You do not need to arrive with perfect clarity.

You only need to start noticing where you keep leaving yourself.

Book an online counselling or performance coaching session with Becoming On Point.

References

Bandura, A. (1977). Self-efficacy: Toward a unifying theory of behavioural change. Psychological Review, 84(2), 191–215.

Bandura, A. (1997). Self-efficacy: The exercise of control. W. H. Freeman.

Carleton, R. N. (2016). Into the unknown: A review and synthesis of contemporary models involving uncertainty. Journal of Anxiety Disorders, 39, 30–43.

Fang, C. M., Liu, A. R., Danry, V., Lee, E., Chan, S. W. T., Pataranutaporn, P., Maes, P., Phang, J., Lampe, M., Ahmad, L., & Agarwal, S. (2025). How AI and human behaviours shape psychosocial effects of chatbot use: A longitudinal randomized controlled study.

Gerlich, M. (2025). AI tools in society: Impacts on cognitive offloading and critical thinking. Societies, 15(1), 6.

Guo, Y. (2026). Meta-cognitive insights into cognitive offloading. Humanities and Social Sciences Communications.

Khosravi, M., & Yoganarasimhan, H. (2026). Impact of AI search summaries on website traffic: Evidence from Google AI Overviews and Wikipedia.

Lazzarelli, A., et al. (2024). Interoceptive ability and emotion regulation in mind-body interventions: An integrative review. Behavioral Sciences, 14(11), 1107.

Reuters. (2026). Young Europeans turn to AI chatbots for emotional support, survey shows.

Risko, E. F., & Gilbert, S. J. (2016). Cognitive offloading. Trends in Cognitive Sciences, 20(9), 676–688.

Sahib, A., et al. (2024). Emotion regulation mediates the relation between intolerance of uncertainty and emotional difficulties. Journal of Affective Disorders.

Suicide Prevention Australia. (2026). Workplace and Industry Distress: Spotlight Report March 2026.

Treves, I. N., et al. (2025). A meta-analysis of the effects of mindfulness meditation on self-reported interoception. Neuroscience & Biobehavioral Reviews.

Van Bael, K., et al. (2024). A systematic review and meta-analysis of the relationship between interoception and alexithymia. Neuroscience & Biobehavioral Reviews.

Wilson, E. J., et al. (2023). The impact of psychological treatment on intolerance of uncertainty: A systematic review and meta-analysis. Clinical Psychology Review.

Zhang, Q., et al. (2025). Generative AI mental health chatbots as therapeutic tools: A systematic review and meta-analysis. Journal of Medical Internet Research.