Scott Adams, creator of the comic strip Dilbert, claims he trained ChatGPT to perform hypnosis, whic
Scott Adams, creator of the comic strip Dilbert, claims he trained ChatGPT to perform hypnosis, which he reportedly used on himself with profound effects. “You have no idea whats coming,” he said on Twitter (aka X) on Sunday.
Adams also recounted his experience on a podcast, describing how he taught the AI language model various persuasion techniques, which he refers to as “waking hypnosis.” This form of hypnosis doesn't involve a trance state but instead uses words to influence individuals in a forceful manner, he said.
Coffee with Scott Adams 7/15/24
— Scott Adams (@ScottAdamsSays) July 15, 2024
“I wondered if I could teach ChatGPT hypnosis, so I taught it some persuasion techniques. I call it ‘waking hypnosis,’” Adams said. Adams, who describes himself as a trained hypnotist and expert persuader on his YouTube channel, shared some of his process.
Adams said he would interrupt the AI's responses to introduce new techniques or refine existing ones. He noted that ChatGPT now remembers information from previous sessions within his account, allowing it to retain the “weapons-grade persuasion” techniques he taught it. In total, he said, his ChatGPT hypnotizer was able to master around six different hypnosis and persuasion techniques before being tested.
“I had the most profound experience,” Adams told his followers. “It gave me an experience that was beyond psychedelics, was basically that powerful, and it did it instantly, and it could do it—it could do it as long as I wanted.”
Adams said his goal was to have AI hypnotize him into a “feel good” state, which, he said, it did. But he refused to share exactly how he did it.
“I know what you're going to ask. You're going to say, ‘What's the prompt? What prompts did you give it to make this happen?’ And the answer is, I will never tell you those prompts,” Adams replied. “They are way too dangerous.”
No, I won't be sharing the prompts. Far too dangerous.
— Scott Adams (@ScottAdamsSays) July 15, 2024
Dr. Pamela Heath—medical doctor, psychologist and hypnotherapist—told Decrypt that self-hypnosis is a legitimate medical technique.
“Hypnosis has been used in medicine for decades, if not centuries,” she said. “I used it in anesthesia and it works.” She also validated the concept of “waking hypnosis,” comparing it to road hypnosis and describing it as “basically boredom induction.”
Regarding the stacking of multiple persuasion techniques, as Adams claimed to have done, Dr. Heath explained that the effectiveness of combined techniques depends on both the hypnotist and the subject. She noted that some skilled practitioners can induce hypnotic states with minimal cues. “Erickson could hypnotize people with a handshake,” she said, referencing renowned psychiatrist and psychologist Milton Erickson. “I've done it with a willing client with simply my consciousness.”
Jack Van Landingham, another certified hypnotherapist and teacher of hypnotherapy at the Center for Advanced Life Skills, said he also trained his own AI chatbot with Ericksons techniques, obtaining “fairly impressive results.”
As a hypnotherapist, and teacher of hypnotherapists, I‘ve experimented with it similarly. After training it in Ericksonian hypnotic language patterns, NLP’s Milton Model, and the use of metaphor, it output some fairly impressive results.
— Jack van Landingham (@JRVanLandingham) July 15, 2024
“I have successfully gotten AI to output, once given the necessary parameters, fairly impressive hypnotic scripts,” he told Decrypt. “Like all AI, it needs to be checked and 'massaged' a bit to fine-tune the result.”
Van Landingham believes that there's an intersection between AI and hypnotherapy beyond simple experiments. “I believe AI can be trained to output hypnotic language that, if read or vocalized by AI, can guide the reader or listener into trance,” he said.
Van Landingham explained that, in his case, he has provided an AI agent examples of issues a person is working on and the model creates a related metaphor using imagery. This, he explains, is a very powerful hypnosis technique.
“Once put in a story format, the person becomes more malleable to changing their mind, literally,” he said.
Scientific studies have shown the potential benefits of self-hypnosis in various areas. A 2020 study found that self-hypnosis could improve sleep quality and duration. Another study from 2021 suggested that hypnosis and self-hypnosis might aid in weight loss when combined with diet and exercise. Additionally, research has indicated potential benefits in chronic pain management, with a study of hospitalized older adults finding that hypnosis offered more pain relief than massage therapy.
It also does not seem to be a harmful technique, as the patient is always in control of their own state—so no, you won‘t be turned into a human zombie for staring at a clock or a black-and-white spiral, and it’s only discouraged in a handful of specific cases.
So why did Adams refuse to share his prompts if they are so beneficial? He warned that once put into motion, the idea of using it “will absolutely take over your brain.”
Dr. Heath echoed this sentiment, warning of potential misuse: “There are unethical people in every field. What happens if a hacker can remotely hypnotize victims to transfer all their money or transfer state secrets or worse?”
Van Landingham also provided an example of how third parties can use hypnosis techniques to harm others.
“This happened to NLP (Neuro-linguistic programming) in the 80s. It started to be used by salespeople to manipulate customers,” he said. “In the 90s, it started to be used by pickup artists to manipulate women. That's how NLP got a negative reputation.”
Added Dr. Heath: “You're wise to get some training in do's and don'ts before trying self-hypnosis.”
Disclaimer:
The views in this article only represent the author's personal views, and do not constitute investment advice on this platform. This platform does not guarantee the accuracy, completeness and timeliness of the information in the article, and will not be liable for any loss caused by the use of or reliance on the information in the article.
0.00