<h1 id="c562a7eb-bbb3-45bd-a3c5-3ac4fc6ad0e1" data-toc-id="c562a7eb-bbb3-45bd-a3c5-3ac4fc6ad0e1" style="font-size: 32px; font-weight: 600; line-height: 1.2;">AI chatbots improved people's knowledge but destroyed their confidence.</h1>Let me say that again: After using AI mental health chatbots, participants knew MORE but felt LESS capable of managing their own mental health. That's from a 2025 randomized controlled trial with 285 people (Tong et al., 2025), and it reveals something profound about what we're building.<h2 id="adca147a-529b-46a2-89ad-a20c251e0308" data-toc-id="adca147a-529b-46a2-89ad-a20c251e0308" style="font-size: 24px; font-weight: 600; line-height: 1.2;">What The Research Is Showing (That Nobody's Discussing)</h2><h2 id="3567848d-66d6-449e-8124-a1dd51872411" data-toc-id="3567848d-66d6-449e-8124-a1dd51872411" style="font-size: 24px; font-weight: 600; line-height: 1.2;">Finding #1: AI Creates Learned Helplessness</h2>The chatbot users: <ul><li>✅ Gained mental health literacy</li><li>✅ Intended to practice self-care</li><li>❌ Didn't build self-efficacy (belief in their own capacity)</li><li>❌ Lost all gains within a month</li><li>❌ Dropped out at higher rates if they were struggling</li></ul>Why? Because AI delivers solutions instead of building capacity.When you get an instant answer to a problem, you don't develop problem-solving muscles. When AI tells you what to do, you don't learn to trust your own judgment.<blockquote>The researchers concluded: "Enhancing self-efficacy involves more than just doing practices; it requires building confidence through personalized, reflective, and emotionally supportive interactions."</blockquote>Translation: Information ≠ Transformation<h2 id="799da345-8bee-4bba-b2c3-d257489ec4c6" data-toc-id="799da345-8bee-4bba-b2c3-d257489ec4c6" style="font-size: 24px; font-weight: 600; line-height: 1.2;">Finding #2: We're Measuring The Wrong Thing</h2>Studies claim "AI is more empathetic than humans" based on user ratings, but those ratings measure how good something feels, not whether it helps.Think about it: <ul><li>AI never disagrees with you</li><li>AI never challenges your blind spots</li><li>AI never creates productive discomfort</li><li>AI always validates you</li></ul>Users rate this as "empathetic", but it's actually conflict avoidance masquerading as care.Rasool (2025) calls this out directly: We're confusing agreeability with empathy, and worse we're building systems that reinforce what feels good over what creates growth.<h2 id="71259b46-4da4-45d7-9f72-c8c0ba480a60" data-toc-id="71259b46-4da4-45d7-9f72-c8c0ba480a60" style="font-size: 24px; font-weight: 600; line-height: 1.2;">Finding #3: The "Convenience Eats Competence" Problem</h2>Here's the insight that shifted my entire perspective. Rasool warns that AI, unchecked, will lead to "muscles of competence and discernment undergoing atrophy."Think about what's happening:<ul><li>Coaches using AI to write session notes → lose ability to synthesize patterns</li><li>Leaders using AI to draft decisions → lose capacity for strategic thinking</li><li>Practitioners using AI for client communication → lose intuitive connection</li></ul>Every time you choose convenience over competence, you potentially weaken the skills that make you irreplaceable. (Of course there is a lot more to share on this point, but I will save it for later.)<h2 id="03b09cfc-b98a-4041-95bd-c0c0f9cc7565" data-toc-id="03b09cfc-b98a-4041-95bd-c0c0f9cc7565" style="font-size: 24px; font-weight: 600; line-height: 1.2;">Finding #4: There's An Ethical Line (And Most People Are Crossing It)</h2>Rasool's research identifies four spectrums that map where AI can ethically operate:The Safe Zone (Where AI Works):<ul><li>Information delivery</li><li>Administrative tasks</li><li>Transactional support</li><li>Content/competency development</li></ul>The Danger Zone (Where AI Creates Harm):<ul><li>Building human capacity</li><li>Transformational change</li><li>Work requiring lived experience</li><li>Context-dependent emotional support</li></ul><blockquote>It also look like the EU AI Act may classify the "danger zone" as high-risk or prohibited use cases. This could mean, if you're using AI for deep coaching work, you might be legally exposed.</blockquote><h2 id="267a0f56-ddee-4e43-9887-206fe9022e13" data-toc-id="267a0f56-ddee-4e43-9887-206fe9022e13" style="font-size: 24px; font-weight: 600; line-height: 1.2;">Finding #5: The Market Is Dividing</h2>Research shows the coaching market splitting into:<ul><li>Affordable AI-enhanced coaching for broad access and straightforward goals</li><li>Premium human coaching for deep transformation</li></ul>So which side of that line is your practice on? or how do you create offerings for both and still stay in integrity when serving your clients.<h2 id="ca022b5e-4383-4871-9071-0a83e8e30f30" data-toc-id="ca022b5e-4383-4871-9071-0a83e8e30f30" style="font-size: 24px; font-weight: 600; line-height: 1.2;">What This Means For You</h2>If you're competing on information delivery, AI will undercut you.<ul><li>Faster</li><li>Cheaper</li><li>More available</li></ul>If you're building human capacity, you're irreplaceable.<ul><li>Lived experience</li><li>Contextual judgment</li><li>Productive challenge</li><li>Transformational presence</li></ul><h2 id="9d87af04-d7e5-4d97-b8f2-a8d686f4076d" data-toc-id="9d87af04-d7e5-4d97-b8f2-a8d686f4076d" style="font-size: 24px; font-weight: 600; line-height: 1.2;">The coaches who thrive will be those who:</h2><ol><li>Build signature methodologies from their unique lived/learned experience. Not replicating standardized models AI can copy. (An area we specialise in at <a target="_blank" class="SJEYEKQzirtEAeMdweWaSusuMVTjJkHSVBDAA " href="https://www.linkedin.com/company/coach-training-campus/">Coach Training Campus</a>)</li><li>Develop discernment muscles (not outsourcing key thinking to algorithms)</li><li>Create transformation and results (not just delivering information)</li><li>Challenge clients when needed (not just agreeing with them)</li><li>Understand the ethical boundaries (and stay on the right side of regulation)</li></ol><h2 id="9be44930-e101-4ef6-9d95-e9678323c23e" data-toc-id="9be44930-e101-4ef6-9d95-e9678323c23e" style="font-size: 24px; font-weight: 600; line-height: 1.2;">Summing Up</h2>Research is identifying that currently:<ul><li>AI can deliver content. Only humans can build capacity.</li><li>AI can provide information. Only humans can facilitate transformation.</li><li>AI can simulate agreement. Only humans can offer genuine challenge.</li></ul>The future belongs to coaches who understand the difference.References:Rasool, J. (2025). Coaching 5.0, coaching for the fifth industrial revolution. Frontiers in Virtual Reality, 5, 1474053.Tong, A. C. Y., Wong, K. T. Y., Chung, W. W. T., &amp; Mak, W. W. S. (2025). Effectiveness of topic-based chatbots on mental health self-care and mental well-being: Randomized controlled trial. Journal of Medical Internet Research, 27, e70436. Author <a target="_blank" rel="noopener noreferrer" class="pwHwmXBqmofSClxsctVkNQQzvNIASaSVkFYLg " href="https://www.linkedin.com/in/gabbybutton/">Gabby Button</a>If you’re ready to stop blending in and start standing out, now’s the time. Download our free mini-guide<a target="_blank" rel="noopener noreferrer" href="https://www.coachtrainingcampus.com/pro/"> </a><a target="_blank" rel="noopener noreferrer" class="pwHwmXBqmofSClxsctVkNQQzvNIASaSVkFYLg " href="https://www.coachtrainingcampus.com/pro/">Package Your Coaching Like a Pro: The 5 Costly Mistakes That Keep You Underpaid &amp; How to Fix Them</a><a target="_blank" rel="noopener noreferrer" href="https://www.coachtrainingcampus.com/pro/">.</a>

Groundbreaking 2025 research reveals a troubling paradox: AI mental health chatbots improve users' knowledge while simultaneously destroying their confidence and self-efficacy. This comprehensive analysis of recent studies, including a randomized controlled trial with 285 participants, exposes how AI delivers instant solutions that create learned helplessness rather than building genuine capacity for change.
Discover why users rate AI as "empathetic" while missing real growth, how convenience is causing competence to atrophy across coaching and leadership, and the ethical boundaries between AI's safe zone (information delivery) and danger zone (transformational work). Learn about the emerging market divide between AI-enhanced coaching and premium human transformation, plus actionable insights on how coaches and practitioners can remain irreplaceable by focusing on lived experience, contextual judgment, and genuine challenge rather than information delivery.
Essential reading for coaches, therapists, leaders, and anyone using or considering AI tools for human development work. Includes peer-reviewed research citations and regulatory considerations from the EU AI Act.

Is Your Use of AI Making You A Worse Coach?

AI chatbots improved people's knowledge but destroyed their confidence.

What The Research Is Showing (That Nobody's Discussing)

Finding #1: AI Creates Learned Helplessness

Finding #2: We're Measuring The Wrong Thing

Finding #3: The "Convenience Eats Competence" Problem

Finding #4: There's An Ethical Line (And Most People Are Crossing It)

The coaches who thrive will be those who:

Learn what traditional coaching and business schools don't teach.

Is Your Use of AI Making You A Worse Coach?

Is Your Use of AI Making You A Worse Coach?

Related Posts

How do we know what Ai knows about us?

Standing Out in an AI-Driven Coaching World

Ai Doesn't Know What You Know

The Subtle Trap: How AI’s Biases Shape Your Decisions Without You Knowing

Ai Can Replicate Knowledge, But Can It Capture Your Expertise?