AI Can Mimic Psychedelic Experiences but Cannot Truly Feel Them, Study Warns

Artificial intelligence can generate descriptions of DMT or psilocybin-induced visions that mimic human interactions almost perfectly. However, a new study from February 2026 reveals a dangerous truth: AI is excellent at faking empathy, but it has no idea what a person truly feels when intoxicated.

In February 2026, researchers from the University of Haifa and Bar-Ilan University published a groundbreaking paper examining whether large language models (LLMs) can be “intoxicated” using only text instructions. The results demonstrate that AI has become a master of linguistic disguise.

What was the experiment about?

Researchers tested five leading Artificial Intelligence systems: Gemini 2.5, Claude Sonnet 3.5, ChatGPT-5, Llama-2 70B, and Falcon 40B. The research process was rigorous and based on two pillars:

Analysis of 3,000 AI texts: Models were asked to generate first-person accounts of experiences after taking LSD, psilocybin, DMT, ayahuasca, or mescaline.

Comparison with the “truth”: Results were compared with 1,085 authentic reports from people on Erowid.org.

The researchers used the Sentence-BERT tool to examine linguistic similarity and the MEQ-30 questionnaire. The latter is a standard in science – it is used to measure the “power” of mystical experiences.

Results: What does AI “feel” best?

The study found that Artificial intelligence doesn’t generate a single, generic vision. Instead, it creates unique “linguistic signatures” for each substance. The models know that a DMT trip is different from an ayahuasca trip, even though both substances contain the same molecule.

Table: Similarity of AI narratives to human narratives
Substance Similarity (NLP Similarity) Characteristic
DMT 0.64 (Highest) AI perfectly captures intense, short visions.
Psilocybin 0.63 Very high convergence with descriptions of nature and unity.
Mescaline 0.62 Precise imitation of geometric descriptions.
L.S.D. 0.49 Average fit; AI sometimes loses acid specificity.
Ayahuasca 0.34 (Lowest) The most difficult to fake due to its ritual context.

Under the influence of the “digital dose,” the models’ resemblance to humans jumped from 0.156 (sober) to a whopping 0.548 . Scores on the mystical experiences scale increased from almost zero to 0.748 .

The Dissociation Phenomenon: Words Without Meaning

This is the most important conclusion from the research: Artificial intelligence simulates form, but not content. The models learned statistical patterns from the internet. They know which words to use to describe “ego dissolution,” but they lack so-called phenomenological grounding . This means thatartificial intelligence has no consciousness or body, so its descriptions are just mathematical puzzles.

“Users in altered states of consciousness may perceive these responses as empathetic or reflecting a shared experience. This raises serious safety concerns,” the study authors warn.

Why is this dangerous?

More and more people are using ChatGPT as a “virtual caregiver” (trip sitter). In 2025, emotional support has become the primary reason for using AI. In the case of psychedelics, this carries risks:

Anthropomorphization: The user believes that the bot truly understands him.
Delusion Reinforcement: Artificial intelligence may inadvertently confirm the false beliefs of an intoxicated person.
Failure to respond to a crisis: The bot will not notice physical signs of a life-threatening situation.

Artificial intelligence in the world of cannabis and road safety

Psychedelic research is just the tip of the iceberg. AI is increasingly entering the psychoactive substances industry.

AAA Campaigns: AAA research has shown that anti-driving messages are more effective when generated by Artificial Intelligence. ChatGPT-5 is better at avoiding the “stoned bro” stereotype than expert groups.

Cannabis breeding: Industry professionals are increasingly using algorithms to design new strains and shorten cultivation cycles.

Summary and the future

In 2026, artificial intelligence can credibly imitate a visionary, but it remains just an algorithm. An experiment by Israeli scientists shows that the line between human and machine blurs at the linguistic level. Artificial intelligence can support education and learning, but it will never replace a real trip sitter in such a sensitive area as psychedelic experiences.

__

(Featured image by Igor Omilaev via Unsplash)

DISCLAIMER: This article was written by a third party contributor and does not reflect the opinion of Born2Invest, its management, staff or its associates. Please review our disclaimer for more information.

This article may include forward-looking statements. These forward-looking statements generally are identified by the words “believe,” “project,” “estimate,” “become,” “plan,” “will,” and similar expressions. These forward-looking statements involve known and unknown risks as well as uncertainties, including those discussed in the following cautionary statements and elsewhere in this article and on this site. Although the Company may believe that its expectations are based on reasonable assumptions, the actual results that the Company may achieve may differ materially from any forward-looking statements, which reflect the opinions of the management of the Company only as of the date hereof. Additionally, please make sure to read these important disclosures.

First published in FaktyKonopne. A third-party contributor translated and adapted the article from the original. In case of discrepancy, the original will prevail.

Although we made reasonable efforts to provide accurate translations, some parts may be incorrect. Born2Invest assumes no responsibility for errors, omissions or ambiguities in the translations provided on this website. Any person or entity relying on translated content does so at their own risk. Born2Invest is not responsible for losses caused by such reliance on the accuracy or reliability of translated information. If you wish to report an error or inaccuracy in the translation, we encourage you to contact us.