Using ChatGPT to design paranormal experiments
I've been poking at the amorphous and questionably sentient ChatGPT chat bot lately and have had good luck working in tandem trying to design new paranormal experiments for fun and zero profit. For those that don't know, ChatGPT is a language model AI that is blazingly fast and can do wild things – synthesize information, make connections, write your damn Catcher in the Rye essay (much to the chagrin of your high school English teacher). You can play with it here
ChatGPT goes beyond just a typical chat bot though — it has the ability to write code for you based on given prompts. For instance, I worked with ChatGPT to build Chant Codes, a simple p5 javascript application that pulls random words out of an English dictionary and displays them randomly on the screen; ChatGPT also gave me sample code to generate random series of tones (click the screen to hear them) and utilizes other visual elements to help emulate the feeling of staring into an analog television set flickering with that nostalgic static 'snow'. I asked ChatGPT to write code that met my specific requirements, asked it to modify bits when it didn't do exactly what I wanted, and then tweaked the final product to build something that kinda... works?
Last night I was talking with ChatGPT about two topics that I always return to: “The Global Consciousness Project” at Princeton and the infamous “Phillip Experiment”. I was surprised that ChatGPT knew about both of these subjects and I set out on a course to figure out how we could use ideas from both of those projects to develop new experiments.
This is how ChatGPT describes the Global Consciousness Project:
“The Global Consciousness Project was an interdisciplinary research project that ran from 1998 to 2018 and aimed to explore the hypothesis that the collective consciousness of humanity can affect physical systems in a measurable way. The project used a network of random number generators (RNGs) located at various locations around the world to attempt to detect changes in randomness that could be associated with large-scale events of collective human consciousness, such as global ceremonies or mass meditations.
And the Phillip Experiment:
The Phillip Experiment was a paranormal investigation conducted by a group of Canadian researchers in the 1970s. The experiment was designed to test the hypothesis that people could create a “ghost” or apparition through repeated visualization and suggestion.The experimenters selected an historical figure, a 17th-century drum major named Phillip, and created a detailed personality and backstory for him. They then attempted to make Phillip “appear” through repeated visualization and suggestion exercises. The experimenters reported seeing and interacting with a “ghostly” entity that seemed to have the characteristics and personality they had assigned to Phillip.
Now how to combine the two? While it nagged about how such an experiment would likely not pass scientific scrutiny (an issue I'm not personally concerned with), it did come up with some ideas for how to move forward:
- Select a historical figure or fictional character with a well-defined personality and backstory.
- Have a group of participants engage in visualization exercises, focusing on the selected figure or character, with the goal of creating a “ghostly” apparition.
- Use GCP-style random number generators to measure any correlations between the participants' visualization exercises and changes in physical data.
- Have participants document their experiences and perceptions during the visualization exercises, including any perceived interactions with the “ghostly” entity.
- Compare the GCP data with the participants' subjective reports to see if there are any patterns or correlations.
- Repeat the experiment multiple times, with different groups of participants, to see if similar results are obtained.
I'm still not fully hip to how these random number generators work, but I am curious about how I might be able to build cheap versions of them. I asked ChatGPT if that was possible, and it gave me insight on some open source projects that utilize cheap hardware to do similar things. The random number generators typically use electronic noise from a diode or other electronic component in order to introduce true randomness, but I asked the AI supposedly it would be possible to use electrical signals generated by the brain, such as EEG (electroencephalogram) signals. That sent me down a whole path to find people who have built DIY EEG devices, including this “build your own mind reading beanie”.
Is this a complete experiment? No. But I appreciate that setting parameters on ChatGPT can get your started on the process to actually design things; while the final product is not always picture perfect, it does interesting things in suggesting correlations, looking for connections and synthesizing data that may be tough to grasp on your own. And now it's got me thinking about building some sort of distributed seance mind-machine that looks for patterns in otherwise statistical randomness while humans try to communicate with Sasquatch. YETI@Home? We'll have to wait and see.
Before I finish, I would like to acknowledge the critiques of these AI applications because of the ethical questions they raise (not to mention how they are likely stealing from artists and other creators and uncompensated labor that feeds their language model). There are also issues with how these companies do content moderation, exploiting cheap labor from Kenyan contractors who had to comb through disturbing images and text and were reportedly not supported psychologically for this traumatic work. You can read more about that here