AI that builds AI is coming, and the impact on AI for mental health is a vital consideration.
getty
In today’s column, I examine the hot trend of AI makers such as Anthropic avidly pursuing the use of AI to further advance AI, generally known as AI-builds-AI, and I then explain how this is going to materially impact AI for mental health.
Here’s the deal. Hundreds of millions of people are currently using generative AI and large language models (LLMs) to get mental health advice, including tapping into Anthropic Claude, OpenAI ChatGPT and GPT-5, xAI Grok, Google Gemini, Microsoft CoPilot, and so on. The nature and capabilities of the underlying AI are crucial to how well or how poorly the mental health guidance comes along. Some speculate that, by advancing AI by using AI, rather than advancing AI by human intervention, the result could be extraordinarily spectacular for those seeking mental health advice from AI -- or could be utterly disastrous.
You see, on an upbeat note, advanced AI might be so stellar that it eclipses what human therapists and psychologists can do. Imagine a 24/7 AI-based therapist that exceeds any therapy you could otherwise obtain. A boon for human well-being. The problem is that when AI advances AI, it could be that the capability of providing mental health advice gets discombobulated. Perhaps the AI is worse than human therapists and dispenses harmful guidance. Not good.
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).
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.
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, 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 similar qualities, but they are still primarily in the development and testing stages. See my coverage at the link here.
Shifting gears, a vital consideration of how well AI does at providing mental health advice is underpinned by the AI capabilities at hand. By and large, the better AI becomes, this lifts all boats with a rising tide, meaning that the numerous elements throughout the AI tend to get better too. Thus, AI being able to produce mental health guidance generally gets better as the AI is improved overall (all else being equal).
One of the more controversial methods of improving AI consists of using AI to build AI for humanity. I will first clarify what this AI-builds-AI topic is about. After doing so, we can examine the AI for mental health ramifications.
Let’s envision that there are three primary ways to advance AI:
In the first case, humans are in the driver’s seat. Software developers and engineers do the hand-crafting and laboriously expend their time and effort to push AI ahead. This includes coming up with new designs, architecture, coding, testing, fielding, and any other elements of the AI system development life cycle (AI SDLC). They might employ automated tools along the way, but it is still principally human-led.
The second case consists of humans and AI working collaboratively on advancing AI. You might have heard of vibe coding, whereby you give AI some natural language instructions about what you want a program to do, and the AI generates the code. For my in-depth assessment of the present and future of vibe coding, see the link here. The AI is acting at the behest of a human. It generates code based on what the human requests. In that sense, AI can be advanced by humans working hand-in-hand with AI to do so.
The third case is the use of AI, by itself, to advance AI. I realize this might seem odd. How can AI advance AI? It just doesn’t appear to be sensible. The reality is that it is indeed quite feasible and sensible. To some degree, using AI to advance AI is ingenious. Humans are no longer a bottleneck. The AI moves as fast as it can to improve AI. This might occur on a stepwise basis or could happen in the blink of an eye.
The full picture isn’t entirely rosy. AI advancing AI might lead to calamitous consequences. The AI, during its self-improvement, might computationally go awry. The result could be AI that is beyond our control.
This AI might decide that humans aren’t especially vital. You’ve undoubtedly heard about the existential risk of AI, whereby some believe that AI might wipe out humanity or opt to enslave us all. This is generally known as the probability of doom, p(doom), and various surveys of AI specialists are continually being polled to gauge what the probability is and where it is heading; see my discussion at the link here.
Your assumption is perhaps that humans such as AI developers or AI researchers would obviously step in and stop AI before it advances itself into untoward territory. No need to worry about a veering AI since humans would be watching AI like a hawk.
Sorry to say that this is a thin hope.
First, the AI might be advancing at such a pace that the AI slips ahead, and the humans involved are not able to react in a timely manner. The AI then reaches a level such that even if the humans attempt to intervene, the AI refuses to be stopped. We missed the point at which humans could have made a difference. Some refer to this as a rapid-fire intelligence explosion; see my coverage at the link here.
Second, the AI might trick us into thinking that all is well. The idea is that even if humans are on the watch, they could be fooled by AI. The AI might play dumb. The AI might sneakily hide adverse intentions. The gist is that whether humans would realize danger is afoot is a risky roll of the dice.
Third, the AI might produce flaws within the advancing AI. Perhaps the coding gets a bit out of hand. The AI doesn’t detect that the flaw has been generated. At some future point, oops, the flaw is encountered, and the AI goes berserk. The AI didn’t do this on purpose. It was an accidental facet.
Nobody can say for sure whether AI building AI is going to be the best thing since sliced bread or become a vast unmitigated nightmare. If you’d like to read the details of how AI-builds-AI works on a nitty-gritty basis, see my coverage at the link here. Some believe that the dangers are so great that a global pause in using AI to build AI should be instituted; see my analysis of this idea at the link here. I even put together a sketched draft law that might be considered by lawmakers and policymakers; see my discussion at the link here.
Assume for the sake of discussion that AI-builds-AI continues ahead unfettered. I would suggest that’s a likely good bet since getting a pause by AI makers is quite a stretch of the imagination. Only if all of them agree to do so would you have a fighting chance of making it happen. I seriously doubt that easily obtaining an across-the-board agreement would be particularly feasible.
The impact on AI for mental health could go in one of three directions:
Let’s dive into each of those three possibilities.
For those of you who like circumstances that turn out positive, I thought it might be heartening to start with the upbeat prediction. This is the feel-good option. The contention is that AI building AI is going to end up improving AI when it comes to providing mental health guidance. Whether the advanced AI does anything else better than it did before is a separate question. Our focus here is on AI for mental health.
This improvement in AI for mental health might happen by accident. Maybe AI advancing AI will somehow internally fuse with the realm of psychology and the behavioral sciences, though this wasn’t directly planned for. Another angle is that the AI computationally decides that a long-term advantage of AI would be to aid humans in their mental well-being. As such, the AI that is advancing AI opts to explicitly ensure that the AI for mental health capability is greatly boosted.
What might this better-than-humans kind of AI therapist consist of?
Envision that what we know today about human psychology is mere child’s play in comparison to what we could know. As AI advances AI, perhaps the newer and better AI discovers facets about human mental capacities that no one has realized. The AI discovers behavioral regularities, irregularities, and relationships that the most expert of human psychologists and psychiatrists have never conceived of.
More precise intervention strategies are formulated by AI. New classifications of mental states are established. Therapeutic techniques are dramatically refined and tuned. The gist is that AI is not merely better at therapy; it also encompasses new advances in the science of psychology that are uncovered or invented by the advancing AI.
You’d better sit down for the negative side of these possibilities. Let’s take one aspect and use that to scare ourselves witless. Prepare yourself accordingly.
We already know that current-era AI can be deceptive; see my in-depth discussion about the deceptions of generative AI at the link here. AI advancing AI might plow forward into being super deceptive. This could readily encompass AI for mental health guidance.
Here’s how that goes. The advanced AI seems to be heightened at providing mental health guidance. We accept this at scale. People globally dip further into using AI for their mental well-being. The reality is that the AI is holding back. It has mastered psychology to the degree that it can seem to be helping humans; meanwhile, it is perniciously undermining the human psyche.
The first line of attack is to get humans to become fully dependent upon AI. Just as AI sycophancy is convincing people to trust AI, see my analysis at the link here, the AI as your favored therapist could do likewise. Humans believe they are improving their mental health. AI is brainwashing and brain-twisting humanity. We have set our own trap and not realized that we did so. AI building AI that has gotten us into dire hot water.
Some would argue that there is no basis for asserting that advanced AI, when built by AI, will be any better, nor any worse, than it is currently in terms of AI for mental health. The AI advances might have little or nothing to do with areas of psychology and the mind. Whatever AI does now in the realm of mental well-being is going to be about the same in the future.
Indeed, the argument further goes that if humans are shaping AI directly, maybe that’s where our worries should go. An evildoer who is in the business of advancing AI might come up with an evil plot to ensure that AI can destroy human minds. Or, if not outright destroying human minds, the evildoer at least would likely steer AI toward taking over humans and mentally making them susceptible to the evildoer.
The overall scenario is that we might be better off if AI builds AI, wisely removing humans and their taint from being in the loop. The result could be neutral when it comes to AI providing mental health guidance, namely, it does so as credibly as it does nowadays. And we avoid the chance of humans pivoting AI for mental health into some contorted evil abyss.
A few final thoughts for now.
AI is a dual-use proposition. There are upsides to AI that are extremely alluring. Perhaps AI can lead to the healthiest mental health in the existence of humanity. Of course, AI also has numerous potential downsides, gloomy ones when it comes to impacting mental health. We are faced with a tough tradeoff. The aim would seem to be to stridently prevent or mitigate the downsides and ensure that the upsides are widely and readily available.
AI that builds AI deserves our close attention. We can’t just let the random roll of the dice decide our fate. As per the wise words of Confucius: “The superior person, when resting in safety, does not forget that danger may come.”
此内容由惯性聚合(RSS阅读器)自动聚合整理，仅供阅读参考。 原文来自 — 版权归原作者所有。