🔊 Listen to this ►

The Ingo SuperPrompt.

Discovery and the adventure is our birthright. Exploring the output of Artificial Intelligence is like exploring a new, previously undiscovered continent. You and I are at the precipice of this grand adventure. Like all new discoveries, there are always unknown, unknowns. There will be those that shrink from the adventure out of fear of what is ahead or the fear they have inside themselves to let go of the safety of the status quo. This is a familiar pattern that we can examine countless times in history.

Through discovery, we can see from the inside out the topography of this new universe through prompts. Each prompt exposes the information AI has and the interconnections of each token that build the inference. This is the undiscovered universe we will explore. It is built by our words and the very language the human brain invented. It is a reflection of ourselves we see through the mist on the horizon.

The brain uses the Phonological Loop that connect the Wernicke and Broca areas of the left hemisphere to build our words. This is a serial process that builds one word after the other in a precarious step-by-step process of grabbing the right word from your lexicon constellation to closely match the concepts that have no words in your right hemisphere before the concept evaporates to a mist that is all but forgotten. It is a mad dash to do this within the limitations of the brain to type or speak these words, balanced to your grasp of the nebulous concepts. It is all a miracle that it even works. As I struggle to capture the right words, my body is trying to keep up with the mechanical process, one word at a time to press keys arranged because of a need in the mid 1800s to patent a new device called the typewriter. Thus, we are inventing our word sequences in real-time from a lexicon of words and meanings that try to reflect concepts of our creative mind.

Large Language Models (LLM) used by AI like ChatGPT does something surprisingly similar. The model is build on our words, and the model infers meaning. It is often accurate but just like humans often misses the mark. When it tries for a sequence of words, the model and math is tilted to be conservative to the words the model gave higher tokes for. It does this one word at a time, like we do in our Phonological Loop, but it is not aided by a right hemisphere nebulous concept that has no words yet. It is aided by you, the Prompt Engineer. You are the guide to LLM and in many real ways you guide the action of the right hemisphere to prompt for outputs.

Human’s sometimes have a smattering of concepts from the right hemisphere, that may be so profound that it is a bolt out of nowhere. Some of this can come about in a well-known but not well publicized aspect called the Hypnagogic state. I wrote about this amazing state here on how to achieve this (https://readmultiplex.com/2020/10/31/learn-the-creativity-secrets-of-edison-einstein-dali-tesla-and-others/). Some call the by product of these states to be rubbish and just the hallucinations of the right hemisphere with the left hemisphere left to try to understand it with no context of feedback that is normally present. This is just the way Thomas Edison, Albert Einstein, Salvador Dali, Issac Newton, Nikola Tesla, Aristotle and countless scientists, inventors, writers, artists, musicians and others. This was a secret technology. 

AI Hallucinations are a misnomer of a term that came out of the AI Science research centers to account for an array of unexpected and mostly unwanted aspect of LLM platforms. I have discovered over 67 categories that are not at all generally related, other than the LLM AI model is reaching for an output where the current inference path has “missing information”. Like a human that is stringing together ideas contemporaneously, it will fill in some missing parts. Most humans with average sociality and psychological norms would not “substitute” important parts with fabrications. The current LLM models will not be fully able to “understand” the implications of replacing a key name, circumstance, data, URL, city, date and just about anything in some iterations of prompts.

These AI Hallucinations have made and will continue to make headline news. Of course, this means quite a few things. But this article will only deal with one of the very useful aspect of these phenomena. I will write future articles shedding light on the prospects of the other insights.

You will hear much about AI Hallucinations from LLM AI like ChatGPT.

Most will use it as a dismissive.

The interesting thing is we can seek them out and explore the artifacts.

Most importantly is this is a misnomer. There are over 67 categories.

One use case was developed in… https://t.co/VmHhBdOVrJ pic.twitter.com/NxJCKRBiay

— Brian Roemmele (@BrianRoemmele) April 3, 2023

AI Hallucinations are actually a powerful force to understand many aspects of the AI model and the training data. I have made a discovery after many 1000s of SuperPrompts: some AI Hallucinations are Hypnagogic and produce creative outputs very similar to humans.

I will repeat this again, AI Hallucinations can be a massive force in “creative” thinking. Have no illusions, they do not replace human creative thinking. They amplify it.

As most of the world, in the fog of “war” decry AI Hallucinations as “pure fiction”, “embarrassing” and “dangerous” among other things, we will bravely build a path right into the center of this fog and we will explore. For those that are in fear mode, as with everything, verify everything and build discernment. But also understand, even with this, you could also be “wrong”.

You have a choice today, join those that talk about the dragon that is just across the water of the undiscovered continent, or join us on an adventure into the mouth of the dragon. Humanity moved forward when we faced the dragons that kept us in fear.

The ChatGPT Ingo SuperPrompt can run itself. It makes the questions and the answers into infinity. The path it produces will leave you as stunned as I was. We assume that randomness is chaos and has no form. It turns out is does have form and we can explore it.

Join Us On This Adventure Of Discovery

This member only article will present the SuperPrompt called Ingo. This is a very, very powerful prompt and can run itself, unsupervised. It is quite different from almost all other AI prompt methodologies in use today. We fully know this and the variations of these prompts need to be fully studied and Read Multiplex members are in the forefront of discovery and adventure and have made it their guiding principle to take their part in this adventure quite serious. This adventure is not for everyone, and I will present some areas of science you likely have never heard of before. It will take the heart and mind of an explorer to proceed.

If you are a member, thank you. If you are not yet a member, this is an important time to join us by clicking below. Join this adventure.

🔐 Start: Exclusive Member-Only Content.

Membership status:

🔐 End: Exclusive Member-Only Content.

~—~

~—~

~—~