A smart way to detox from overreliance on AI in mental health guidance.
getty
Today’s column looks at the growing need for some people to undergo a kind of detox regarding the overuse of AI as mental health advisors.
This is the contract. People are increasingly relying on generative AI and large-scale language models (LLMs) for ongoing mental health. They log into popular LLMs like ChatGPT, Claude, Gemini, and Grok and regularly ask questions about how to address mental health concerns they have. This is more than just looking up information about mental health conditions. Some people’s aim is to rely heavily on AI as a kind of therapist, using AI to assess an individual’s needs and provide professional-quality psychological advice on par with a human therapist.
General-purpose generative AI does not yet provide that level of therapeutic functionality. There are customized LLMs that are trying to get into that space, but even they are still not a solution that guarantees bypassing human therapists altogether. Generally, these specialized LLMs are used in conjunction with a therapist in a way that I have described as a new triad. This is a therapist-AI-client relationship that goes beyond the traditional therapist-client dyad (see my discussion at the link here).
The bottom line is that people are using AI as mental health advisors far beyond what modern LLMs can achieve. And we often become overly reliant on AI. There’s not a day that goes by that I don’t log in to AI and consult with someone. Their excesses are likely to do more harm than good. Somehow we need to find a way to move away from over-reliance on AI.
There are smart ways to break that addiction.
Let’s talk about it.
This analysis of AI breakthroughs is part of my ongoing Forbes column on emerging AI, which includes identifying and explaining the various complexities of influential AI (see link here).
AI and mental health
As a quick background, I have extensively covered and analyzed myriad aspects of the emergence of modern AI to generate mental health advice and implement AI-driven treatments. This increased use of AI is primarily driven by the evolution and proliferation of generative AI. For a quick overview of some of the columns I’ve posted on this evolving topic, see the link here. This is a quick summary of about 40 of the 100+ columns I’ve posted on this subject.
There is little doubt that this is a rapidly evolving field with significant benefits, but at the same time, unfortunately, these efforts also have hidden risks and obvious pitfalls. I speak out frequently about these pressing issues, including appearing on an episode of CBS last year. 60 minutessee this link.
Background on AI for mental health
I hope to lay the groundwork for how generative AI and large-scale language models (LLMs) are typically used in ad hoc ways for mental health guidance. Millions of people are using generated AI as an ongoing advisor on mental health considerations (note that ChatGPT alone has over 800 million weekly active users, a significant proportion of whom dip into mental health aspects; see my analysis at the link here). The most common use of modern generative AI and LLM is to consult the AI on aspects of mental health. Check out my article at the link here.
This common usage makes a lot of sense. You can access most of the leading generative AI systems anytime, anywhere, at near-free or very low cost. So if you have any mental health concerns and would like to discuss something, just log in to AI and talk to us right away, 24 hours a day, 7 days a week.
There are major concerns that AI could easily go erratic and provide inappropriate or grossly inappropriate mental health advice.
Banner headlines this August reported on a lawsuit filed against OpenAI for a lack of AI safeguards when it comes to providing cognitive advice. Although AI manufacturers claim to be phasing in AI safety measures, there are still many downside risks from AI doing untoward things, such as helping users unknowingly co-create delusions that could lead to self-harm.
For more information on the OpenAI lawsuit and how AI fosters paranoid thinking in humans, please see my analysis at this link. I have been enthusiastically predicting that eventually all major AI manufacturers will be taken to the woodshed due to lack of robust AI safeguards.
overuse is occurring
In some ways, when it comes to the emergence of generative AI and its use in mental health, some may cite the old saying that you can never have too much of a good thing.
It probably makes sense to leverage AI for occasional questions about mental health. However, it is important to be aware that AI is not equivalent to a human therapist. Additionally, it is important to remain skeptical and not blindly accept AI-generated mental health guidance.
At some point, people will start focusing on leveraging AI as a mental health advisor. A common sign is that people very often consult AI about mental health issues and allow AI to clearly guide their mental health. They fail to strike the right balance in how they use AI for this important purpose.
I tend to layer this type of usage into five progressive levels:
- level 1: Don’t overuse AI for mental health.
- level 2: Just a little overuse of AI for mental health.
- level 3: Moderately overusing AI for mental health.
- level 4: It is clear that we are overusing AI for mental health.
- level 5: Overusing AI for mental health.
If you find yourself at level 3 or above, you may want to reconsider what you are doing. Once you enter level 3, there’s a bit of a slippery slope that quickly spirals up into levels 4 and 5. This can happen without you even realizing it. People suddenly find themselves incredibly dependent on AI for their mental health.
Also, keep this thorny topic in mind for others you may know who may be on a similar path of inadequacy. If you suspect that they are inexorably making themselves worse by relying too heavily on AI for their mental health, find out what you can do to help them before they go too far down the rabbit hole.
Steps to detox
Suppose someone is a mental health counselor and wants to break free from their dependence on AI. We’ve put together six basic steps that you should consider carefully.
The six steps are:
- (1) Consult a human therapist.
- (2) Taper off gradually instead of going cold turkey.
- (3) Rebuild human support channels.
- (4) Develop alternative coping skills.
- (5) Reimagine the role of AI in your life.
- (6) instruct the AI to moderate or cease providing such advice;
Depending on the severity of the overreach, it may be wise to start with the first step: consulting a human therapist. For milder conditions, you can proceed with the steps in any order. Also, you don’t have to be perfect at each step. If we do what we can, our reliance on AI in mental health could weaken.
Let’s briefly explain each of the six steps.
Consult a human therapist
Those who rely too heavily on AI for their mental health would undoubtedly benefit from consulting a human therapist about this issue.
Addicted people are also more likely to not only need help to stop using AI, but also have other mental health conditions that a human therapist can detect. Dependence on AI is perhaps one of the mental health issues of the moment.
Turkey will shrink more gradually than it will cool down
One seemingly obvious approach is to stop using AI for mental health purposes without thinking.
That may work, but the possible negative effect is that this could be significantly destabilizing and potentially unsettling. Gradual tapering may provide a smoother way to stop using AI. For example, reduce the use of AI from 7 days a week to 3 days a week. Then reduce it to once a week. This may involve setting time limits. Limit screen time so each interaction becomes progressively shorter.
Rebuild human support channels
Over-reliance on AI for mental health advice often leads people to ignore human support channels and focus all their attention on AI. They abandon human interaction and lose relationships.
The human-first rule could be a useful mechanism to reduce reliance on AI. Can I rely on family or friends? Do you have a mentor you can talk to? This can help restore balance to your being. AI usage is reduced. Human relationships will be enriched.
Develop alternative coping skills
Rather than relying on AI as a coping mechanism, try to develop alternative coping skills.
Can you enhance your own decision-making and self-reflection? Is there a guidebook to resist the urge to consult an AI? The hope is that you will become more independent and able to take charge of your own mental health care.
Reframe the role of AI in your life
There is an unfortunate ambiguity between perceiving AI as a friend rather than a reference tool (see my analysis of AI peer relationships linked here).
Once someone goes down the path of friendship, they begin to attribute a sense of omniscience, a sense of intimacy, and other anthropomorphic labels to the AI. Try reframing the nature of AI and what it is. AI is a computational and mathematical function that can be used as a reference tool. AI doesn’t “know you” or “understand you.” Don’t let your emotions override your good judgment.
Tell the AI to moderate such advice or stop providing such advice
You can involve AI in the detoxification process.
Enter a prompt to tell the AI to stop giving mental health advice. These can be made into system commands and applied to all AI chats. Again, you can also do this in stages, with prompts that are less permissive about mental health guidance at first. The downside to this approach is that the AI may temporarily stop acting and then go back to its old ways, or it may incorrectly calculate that your protest is an invitation and thus become computationally more assertive regarding mental health issues.
detox battle
These are just a few of the detoxification pathways.
Overall, always keep your eyes wide open when it comes to using AI for mental health. Dipping your toe in it is usually fine. Jumping in with both feet is a somewhat disconcerting approach. Another angle is to avoid the drips and drips of being slowly but inevitably drawn into the control of the AI. If you’re worried that if you’re even remotely interested in using AI for this vaunted purpose, you’ll fall in love with it, then you’re probably better off not doing it at all.
Final thoughts for now.
Famous novelist Raymond Chandler insightfully said, “There is no more deadly trap than the one you set for yourself.” The ubiquitous nature of generative AI and the free or easy means of accessing it is a trap waiting to be set. Relying too heavily on AI for your mental health can be like walking through quicksand without realizing it.
Be cautious and remove yourself from the mud as soon as possible if necessary.
