💾 Archived View for paraweird.org › news › 2023-01-30-chatgpt.gmi captured on 2024-05-10 at 10:43:58. Gemini links have been rewritten to link to archived content
⬅️ Previous capture (2023-03-20)
-=-=-=-=-=-=-
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).
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 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.
"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.
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:
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 to monkey around.
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 and 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, for instance:
A "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 were paranormal experiments; 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.
https://mashable.com/article/chat-gpt-open-ai-workers-exploitation
Mashable Article on Chat GPT And Exploitation