By David Stephen
There is a new [March, 2026] report from United for Medical Research, NIH’s Role in Sustaining the U.S. Economy, stating that, “In Fiscal Year 2025, the National Institutes of Health (NIH) awarded $36.58 billion in research grants to organizations in the 50 U.S. states and the District of Columbia.”
“According to UMR’s analysis, this research funding supported 390,863 jobs and produced $94.15 billion in new economic activity nationwide in 2025. For every $1 invested in NIH research, there was $2.57 generated in economic activity — a 250% return.”
“Although total funding remained strong, FY2025 saw significant shifts in how awards were distributed. NIH made expanded use of multi-year funding (MYF) — obligating the full value of certain grants upfront rather than distributing funding annually.”
“While this approach enabled NIH to spend its full budget, it resulted in 5,564 fewer grants funded in FY2025 compared to FY2024 since money set aside for the remaining years of the MYF grants was not available to support other new and competing research projects.”
AI at the NIH
The NIH should be thoroughly pursuing investments in artificial intelligence, related to health solutions at this time, with ballistic aggression and desperation to get into solving for areas of vulnerabilities that the major AI companies have deserted. Simply, the path to supremacy for the NIH, in the AI era, is connected to seeking out answers, for side-effects of consumer AI, at a scale that would be presently dominant, as well as fit into future adoption.
There is possibility for the NIH to make some investments in solutions for health AI, that could return so much, that for future budget cuts, the NIH may close in on self-sustenance.
For example, the biggest health problem in AI for now is AI psychosis and delusion, where some users have delusions [induced or reinforced], with others driven to psychosis, self-harm or worse.
There is no startup in the world that is taking on this problem directly for now. There is also no AI psychosis research lab. There is no product from any company that is looking to solve this problem. There is no angel investor, institutional investor, venture capital, private equity, or asset management that has considered to explore this problem towards a relevant and then profitable solution.
This makes it an opportunity for the National Institutes of Health, in a way to also be at the fore of preventive health for consumer AI, even as personal use of AI grows, for therapy, companionship, relationships, friendships and much else.
Solving AI Psychosis
The solution to AI psychosis is to provide a dynamic display of the human mind, presenting it as a flowchart. The solution will assume the mind has destinations and relays between them.
Then, whatever the themes are — or the keywords of a chat, with an AI chatbot — will play as going to some destinations, in the mind and ignoring others. These destinations could be pleasure, delight, grandeur, care, support, and so forth. While destinations ignored could be caution, consequences, fear, doubt and so forth.
Also, the relays to the destinations may not use those for reality, but could be those for non-reality. Simply, in the mind [conceptually], there are relays for reality, such as interpretations of external sensory inputs. There are also those of non-reality like imagination, fantasy, and so forth.
If it happens that the route for non-reality is getting used for things in reality, there could be breaks from reality, resulting in problems. Showing these as a dynamic display, as well as resulting a [mind safety] score could become a major solution [that is applicable] across platforms, as AI psychosis risks mount.
It can also be useful against social media addiction, gambling addiction and video games addiction. It will be a separate application, but the API will be available for all platforms.
The NIH Brain Initiative has an opportunity to get ahead of this, just like any angel investor, venture capital, private equity or asset management, that sees profitability in an unexplored and scalable AI area, since there’s already a robust architecture.
The solution is based on the postulation in Conceptual Biomarkers and Theoretical Biological Factors for Psychiatric and Intelligence Nosology.
This solution will be subscription only, with free versions for general displays without specificity. It is possible to have the product ready by April 10, 2026, if the startup is incorporated this March or if the product is subsumed within some existing portfolio.
There is a new [March, 2026] paper in The Lancet Psychiatry, Artificial intelligence-associated delusions and large language models: risks, mechanisms of delusion co-creation, and safeguarding strategies, stating that, “Large language models (LLMs) are poised to become a ubiquitous feature of everyday life, mediating communication, decision making, and information curation across nearly every domain. Within psychiatry and psychology, the attention has largely been on bespoke therapeutic applications, sometimes narrowly focused and often diagnostically siloed, rather than on the broader reality that individuals with mental illness will increasingly engage in agential interactions with artificial intelligence (AI) systems. Although the capacity of these systems to model therapeutic dialogue, provide companionship at any hour of the day, and assist with cognitive support has sparked understandable enthusiasm, these same systems might contribute to the onset or exacerbation of psychotic symptoms. Emerging evidence indicates that agential AI might validate or amplify delusional or grandiose content, particularly in users already vulnerable to psychosis, although it is not clear whether these interactions can result in the emergence of de novo psychosis in the absence of pre-existing vulnerability. Some individuals might benefit from AI interactions, for example, where the AI agent functions as a benign and predictable conversational anchor, but there is a growing concern that these agents could reinforce epistemic instability and blur reality boundaries. In this Personal View, we outline the emerging risks, possible mechanisms of delusion co-creation, and safeguarding strategies for agential AI for people with psychotic disorders. We propose a framework of AI-informed care, involving personalised instruction protocols, reflective check-ins, digital advance statements, and escalation safeguards to support epistemic security in vulnerable users. These tools reframe the AI agent as an epistemic ally (as opposed to a therapist or a friend), which functions as a partner in relapse prevention and cognitive containment. Given the rapid adoption of LLMs across all domains of digital life, these protocols must be urgently co-designed with service users and clinicians and tested in clinical trials.”
