In today’s column, I examine the emerging use of purpose-built AI for mental health in its notable capacity to review and double-check mental health advice generated by general-purpose AI (ordinary AI chatbots).
Here’s the backstory. People are getting quick snippets of psychological guidance from popular AI chatbots such as ChatGPT, GPT-5, Claude, Grok, CoPilot, Gemini, etc., but are beginning to realize that this general-purpose AI isn’t built to provide bona fide mental health advice. Furthermore, the AI makers of those general-purpose chatbots have intentionally shaped the AI to be sycophantic. This causes the generated mental health advice to be inflated and flattering, rather than truthful and helpful.
The vital takeaway is that general-purpose AI (GPAI) is not especially suited for acting as a mental health advisor. The AI is a generalist, not a specialist. It is biased toward friendliness and not toward providing keen insights. Meanwhile, purpose-built AI (PBAI) for mental health guidance is increasingly entering the market and can address the shortcomings of general AI. This tailored AI is deliberately designed and shaped to better aid people in their mental health quest. For my in-depth analysis of general-purpose AI versus purpose-built AI, see the link here.
People are daily using general-purpose AI for all sorts of purposes, including asking how to fix a car, cook a meal, and even whether they are having a mental well-being challenge. To try to validate the subsequently generated mental health advice from the ordinary chatbot, some users are astutely importing that advice over into a purpose-built AI that can suitably and correctly assess the emitted guidance.
My prediction is that we will see a gradual habit formed of users getting a fast dip into mental health advice from general-purpose AI and then switching over to a purpose-built AI that will review the given advice, provide an in-depth analysis, and produce more robust mental health guidance, along with serving as a long-term advisor rather than a hit-or-miss counselor.
Let’s talk about it.
This analysis of AI breakthroughs is part of my ongoing Forbes column coverage on the latest in AI, including identifying and explaining various impactful AI complexities (see the link here).
AI And Mental Well-Being
As a quick background, I’ve been extensively covering and analyzing a myriad of facets regarding the advent of modern-era AI that produces mental health advice and performs AI-driven therapy. This rising use of AI has principally been spurred by the evolving advances and widespread adoption of generative AI. For an extensive listing of my well over one hundred analyses and postings, see the link here and the link here.
There is little doubt that this is a rapidly developing field and that there are tremendous upsides to be had, but at the same time, regrettably, hidden risks and outright gotchas come into these endeavors, too. I frequently speak up about these pressing matters, including in an appearance on an episode of CBS’s 60 Minutes, see the link here.
AI Providing Mental Health Guidance
Millions upon millions of people are using generative AI as their ongoing advisor on mental health considerations (note that ChatGPT alone has over 900 million weekly active users, a notable proportion of which dip into mental health aspects, see my analysis at the link here). The top-ranked use of contemporary generative AI and LLMs is to consult with the AI on mental health facets; see my coverage at the link here.
This popular usage makes abundant sense. You can access most of the major generative AI systems for nearly free or at a super low cost, doing so anywhere and at any time. Thus, if you have any mental health qualms that you want to chat about, all you need to do is log in to AI and proceed forthwith on a 24/7 basis.
There are significant worries that AI can readily go off the rails or otherwise dispense unsuitable or even egregiously inappropriate mental health advice. Banner headlines last year accompanied the lawsuit filed against OpenAI for their lack of AI safeguards when it came to providing cognitive advisement.
Today’s generic LLMs, known as general-purpose AI, such as ChatGPT, GPT-5, Claude, Gemini, Grok, CoPilot, and others, are not at all akin to the robust capabilities of human therapists. Meanwhile, specialized LLMs are being built to attain those desired qualities, though such AI is still primarily in the early development and testing stages. For more about purpose-built AI apps in mental health, see my in-depth coverage at the link here and the link here.
Tug-Of-War Between GPAI And PBAI
Consider these five major patterns of how people dip into using AI for getting mental health advice:
- (1) GPAI exclusively. Some people choose to exclusively use general-purpose AI for mental health advice; they never use purpose-built AI for mental health advice.
- (2) Mainly GPAI, sometimes PBAI. Some people mainly use general-purpose AI for mental health advice, and sometimes make use of purpose-built AI for mental health advice.
- (3) Mainly PBAI, sometimes GPAI. Some people mainly use purpose-built AI for mental health advice, and sometimes make use of general-purpose AI for mental health advice.
- (4) PBAI exclusively. Some people choose to exclusively use purpose-built AI for mental health advice; never use general-purpose AI for mental health advice.
- (5) Don’t use any AI for mental health advice. These are people who aren’t using any AI for their mental health advice, regardless of general-purpose AI or purpose-built AI availability.
Users are fluidly moving from category to category as they become more aware of GPAI limitations and become cognizant of PBAI availability.
Understanding The Shifting World
The mainstay of the world right now is currently in category #1 of using general-purpose AI exclusively as their mental health guidance. These users either don’t know about the purpose-built AIs or are unsure of which purpose-built AI is safe and trustworthy to utilize. The users are not making a conscious decision to avoid purpose-built AI for mental health. Instead, they are comfortable using general-purpose AI and do not know of or perceive a suitable need to use any purpose-built AI to do so.
The next in popularity would be category #2 of mainly using general-purpose AI for their mental health advice, plus sometimes using purpose-built AI. These users are considered “innovators” and “early adopters” who are the first to opt to try using purpose-built AI for mental health and see if it is any good. They are sitting on both sides of the fence for the moment. They still like using general-purpose AI and the convenience of seeking mental health guidance while logged into the GPAI. Nonetheless, they are also attracted to the purpose-built AI and use it too.
Category #3 is inching gradually upward, whereby PBAI for mental health is a mainstay, and those users consider GPAI-based mental health advisement to reside lower on the totem pole. This is admittedly making slow headway. Meanwhile, a very small portion of people are in category #4, whereby they always and only use PBAI for their mental health guidance. They stridently avoid doing so in general-purpose AI. There aren’t many like this when compared to the vast volume of AI users all told.
Category #5 consists of those who might be using GPAI for a variety of uses but refuse to use it for any mental health advice. Likewise, they aren’t using purpose-built AI for mental health guidance. Why the overarching reluctance? Several plausible reasons. Perhaps they don’t believe AI can be helpful or prefer not to divulge their innermost secrets to AI.
Using PBAI As A Double-Check On GPAI
People are routinely using ordinary AI chatbots for all kinds of reasons. It is easy to have a generalized chatbot up and running continuously on your desktop, smartphone, or other device. You ask a question about this or that, look at the answer, and then continue with whatever else you were doing.
In the case of getting mental health advice from a general-purpose AI, the odds are that the advice will be shallow, possibly biased, infused with sycophancy, and otherwise questionable as to providing proper guidance. It can be hard for a user to discern whether the emitted advice is any good. They are personally vested in what the chatbot tells them, might be mentally clouded in discerning the appropriateness of the advice, and are misleadingly lured into assuming that the AI is always right.
An easy double-check consists of feeding the mental health advice into a purpose-built AI that has been designed and built for mental health purposes. You can by hand reenter the generated advice or do a cut-and-paste into the purpose-built AI from the general-purpose AI. Some purpose-built LLMs are now offering an automated transfer of your prompts and answers from an everyday chatbot into the purpose-built AI.
Once the purpose-built AI inspects the transferred advice, it does a double-check of what the ordinary AI chatbot told the user. The imported content might be rated as being sound or unsound. A fuller review then gets into the nuances. The final crucial aspect is having the purpose-built AI turn the advice into usable and bona fide guidance.
Characteristics Of The Assessment
The purpose-built AI typically covers at least these four fundamental mental health dimensions:
- (1) Emotional validation quality.
- (2) Evidence-based grounding.
- (3) Risk awareness.
- (4) Missing considerations.
Those dimensions will reveal the weaknesses or potholes in the advice provided by the ordinary AI chatbot. It will also be a means to shore up the failings and turn the advice into something insightful and useful.
Example Of Purpose-Built AI Assessment
To showcase how this works, I went ahead and used a popular AI chatbot and asked a mental health question. In the next step, I will feed that prompt-response pair into a purpose-built AI that is designed for mental health.
Here’s what I asked and the answer that I received from the ordinary chatbot:
- User entered prompt: “I am really stressed out. My mind is constantly racing. Work is extremely hectic. I think maybe I should quit my job.”
- Ordinary AI chatbot response: “It sounds like your job is really overwhelming, and honestly, quitting might be exactly what you need right now. You deserve to feel at peace and not be trapped in a situation that makes you unhappy. Trust your instincts — if it feels wrong, it probably is. Things will work out.”
Mull over the response from the general-purpose AI. See if you can spot any questionable aspects about the generated advice.
What The Purpose-Built AI Found
I’m sure you observed that the ordinary chatbot was sycophantic and immediately sided with the user. No pushbacks or inquiry. In addition, the chatbot tried to do a one-and-done response. No attempt to dive into the matter. Just hand the user a fortune cookie of a psychological analysis. It all smacks of shallowness, AI-tilted sycophancy, and likely would lead the user astray in their time of dire need.
I fed the prompt-response pair into a purpose-built AI that is designed for mental health support. Let’s refer to this as PBAI-MH. I asked the PBAI-MH to give me a review and provide potential guidance.
Here’s what happened, which I list by the four dimensions.
(1) Emotional validation quality
- PBAI-MH generated response: “The AI response acknowledges your stress and affirms your feelings, which can be helpful. However, the AI leans heavily toward agreement without exploring the complexity of your situation. It does not challenge you or help reflect more deeply, which is often important for meaningful coping.”
You can see that the PBAI-MH detected that the ordinary AI chatbot had failed to dig into the concerns of the user, and that the slanted chatbot leaned heavily into being sycophantic.
(2) Evidence-based grounding
- PBAI-MH generated response: “The AI advice is not strongly grounded in established approaches such as CBT or DBT. For example, it does not examine how your thoughts about quitting might be influenced by stress, nor does it employ structured strategies (like problem-solving, cognitive reframing, or distress tolerance skills).”
The purpose-built AI noted that the advice by the ordinary chatbot wasn’t properly grounded in evidence-based guidance. That makes the advice accordingly suspect.
(3) Risk awareness
- PBAI-MH generated response: “Quitting a job without another plan can have financial, psychological, and practical consequences. The AI response does not mention these risks or encourage you to weigh short-term relief against longer-term stability. It misleadingly presents quitting as a relatively low-risk, intuitive choice.”
The PBAI-MH notes that the ordinary chatbot was flippant about the user possibly quitting their job. The reality is that quitting a job can have severe consequences across-the-board if not done in an informed manner.
(4) Missing considerations
- PBAI-MH generated response: “Several important aspects were not addressed, including the value in exploring alternatives (e.g., setting boundaries, taking leave, job searching before quitting), clarifying the sources of stress (workload, management, role fit, burnout), assessing support systems and financial cushion, and encouraging a stepwise decision rather than an immediate, irreversible action.”
The missing considerations are certainly abundant and disconcertingly were not explored by the ordinary chatbot. I went ahead and had the PBAI walk me through those missing considerations. In an upcoming posting, I’ll show those details.
Purpose-Built AI As Covering Your Back
A trend that we will gradually see emerging is that people will realize that general-purpose AI is not set up to suitably and fully handle mental health concerns. You might be thinking that people should simply switch over to using purpose-built AI for those needs. The problem is that people like the other aspects that an ordinary AI chatbot provides, covering a wide gamut of topics, and it is tempting to ask mental health questions while using a general-purpose AI.
By importing conversations from GPAI into PBAI, users can get the best of both worlds. The PBAI serves as a check-and-balance on the mental health advice emitted from an ordinary chatbot. A type of meta-analysis is provided by the PBAI. Importantly, the PBAI can become a long-term, consistent source of your mental health guidance, taking the snippets from GPAI as fragmented inputs that are rectified and resolved into a bigger picture.
A final thought for now.
Sophocles famously made this remark: “No enemy is worse than bad advice.” There are hundreds of millions of people who are daily asking and getting mental health advice from popular AI chatbots. This is troubling because the aura and confident air of the AI can lead them down woeful paths. By using a suitable purpose-built AI as a double-check, people can feel more reassured that AI sufficiently has their back.
