Personal Perspective: How I learned about the associative power of AI. #
Posted June 2, 2026 [ Reviewed by Monica Vilhauer Ph.D.
](/us/docs/editorial-process)
Key points
- AI input led to development of a hypothesis that spiritual practice may help improve OCD.
- AI's ability to synthesize data from diverse databases has led to scientific discoveries.
- Interactions with AI can help generate hypotheses that researchers can investigate.
I have been using Claude (an AI) to help research information regarding a new book that I am writing. As part of the feedback that Claude has given me, it has presented unexpected combinations of ideas derived from the material that I have been researching. These combinations were made possible because of its ability to rapidly synthesize information from diverse databases.
Some of these combinations have helped me discover new intriguing ways of thinking about the subject matter at hand. Because of this, I realized that AI may help accelerate the occurrence of serendipitous scientific discoveries.
The Discovery of Antibiotics #
Major advances in science initially were based on observations of events around us. A famous example of a serendipitous observation involved Dr. Alexander Fleming’s discovery that led to characterization of antibiotics.
Fleming was growing bacteria in petri dishes and noticed that in one dish, no bacteria grew in an area of mold contamination. He asked himself how this could be, and reasoned that the mold produced an anti-bacterial compound. Fleming's observation eventually led to the development of penicillin, one of the most transformative medical discoveries of the twentieth century (Lax, 2004).
A Patient With PANDAS and OCD #
The same spirit of curiosity that drove Fleming to ask how rather than simply move on led me to look more closely at a remarkable patient I describe in my book, who dealt with Pediatric Autoimmune Neuropsychiatric Disorders associated with strep infections (PANDAS). With this illness, it is thought that the inflammation produced by the body to fight the strep bacteria ends up attacking and damaging the brain.
Because of his illness that occurred when he was 11 years old, my patient developed rapid onset anxiety and obsessive-compulsive disorder (OCD) so severe that he could not leave his room without performing a long series of rituals that were difficult to complete. As a result, he rarely left his room for four years, did not attend school, and lost his friends.
My patient improved greatly with the use of self-hypnosis for regulation of his OCD, and with hypnosis-facilitated guidance from his subconscious. Within four months he was able to leave his room, re-establish social connections, and return to school.
He then began expressing deeply spiritual thoughts that arose from exploration of his subconscious realm. Further, he became engaged in daily sessions of self-reflection that strengthened his spiritual beliefs.
Rituals in PANDAS, OCD, and Religious Practice #
I discussed with my patient that there is a notable overlap between rituals we observe in OCD and rituals in religions, as Freud (1907) initially observed. For instance, both religious rituals and OCD compulsions require exact repetitions of specific actions. If an error is made, the ritual needs to be restarted.
Rituals involving purification and cleansing, which carry deep spiritual significance in Judaism, Islam, and Hindu practice, share structural similarities with OCD contamination fears. Further, when people experience unwanted, sinful thoughts, Judaism, Christianity, and Islam have specific rituals and theological frameworks to deal with these.
There is even a sub-type of OCD, scrupulosity, which involves obsessive and compulsive thoughts of a religious nature, such as a fear of having sinned, distress because of blasphemous thoughts, compulsive confessing, and compulsive reading of religious texts (Nelson et al., 2006). Sometimes, it is difficult to distinguish between OCD and sincere religious fervor.
As I was discussing rituals with Claude the AI, it pointed out that religious rituals (Graybiel, 2008), OCD (Graybiel & Rauch, 2000), and PANDAS (Swedo et al., 2004) are all thought to be related to activity or dysregulation in the brain’s basal ganglia and provided some citations demonstrating this observation. (I verified the accuracy of these citations in Pubmed, a bibliographic database, given the propensity for AI to hallucinate.) I reflected at that point that AI excels at pointing out associations derived from its very broad base of information.
Given Claude’s observation, I conjectured that my patient’s development of spirituality as he recovered might have been directly associated with the changes in his brain caused by PANDAS. This hypothesis is supported by my observation that several other patients in my clinical practice with a history of PANDAS and OCD also have become very spiritual.
Scientific studies will need to be performed to assess the merit of my hypothesis, but I found it notable that it was the information pointed out to me by an AI that prompted me to form this new idea.
Claude again demonstrated its associative capacity when it pointed out that OCD is characterized by intrusive thoughts and compulsive attempts at control in contrast with spiritual practice centered on releasing control, trusting a higher power, and accepting what is received without judgment.
This helped me recognize that spiritual practice could be therapeutic for OCD, as reflected by research that has demonstrated the benefit of mindfulness and acceptance-based programs in the treatment of OCD (Bürkle et al., 2025).
AI Can Accelerate Scientific Progress #
My recent experiences with Claude led me to explore how AI already has accelerated scientific progress. I found out that analogous to how Fleming’s unexpected observation led to the development of antibiotics, in 2020, an AI analysis of the molecular structure of over 100 million chemical compounds led to the surprising finding that a drug originally investigated as a treatment for diabetes could function as a powerful antibiotic (Stokes et al., 2020).
Takeaway #
Our use of AI is at a relatively early stage. Thus, I think that as AI is further refined (including beyond our current use of large language models that drive AIs such as Claude), many more scientists will routinely utilize cutting-edge AI technology to inform and accelerate their research progress.
For instance, the hypothesis about my patient's spiritual development and basal ganglia changes needs to be tested, and this is exactly how AI-accelerated discovery should work: It can help generate hypotheses that researchers can pursue. References
Bürkle J. J., et al.. (2025). Mindfulness- and acceptance-based programmes for obsessive-compulsive disorder: A systematic review and meta-analysis. J Anxiety Disord. Epub 2025 Jan 17.
Freud, S. (1907). Obsessive Actions and Religious Practices." Imago, Vol. 1.
Graybiel, A. M. (2008). Habits, Rituals, and the Evaluative Brain. Annual Review of Neuroscience, 31, 359–387
Graybiel, A. M., & Rauch, S. L. (2000). Toward a neurobiology of obsessive-compulsive disorder. Neuron, 28(2), 343–347.
Lax, E. (2004). The Mold in Dr. Florey's Coat: The Story of the Penicillin Miracle. Henry Holt.
Nelson, E. A., et al. (2006). Scrupulosity in Patients with Obsessive-Compulsive Disorder: Relationship to Clinical and Cognitive Phenomena." Journal of Anxiety Disorders, Vol. 20.
Stokes, J. M., et al. (2020). A deep learning approach to antibiotic discovery. Cell, 180(4), 688–702.
Swedo, S., et al. Pediatric Autoimmune Neuropsychiatric Disorders Associated with Streptococcal Infections (PANDAS) 2024 Jul 7 [Updated 2024 Sep 13]. In: Ferretti JJ, Stevens DL, Fischetti VA, editors. Streptococcus pyogenes: Basic Biology to Clinical Manifestations [Internet]. 2nd edition. Oklahoma City (OK): University of Oklahoma Health Sciences Center; 2022 Oct 8. Chapter 26.