DOCS For The Awesome.AI Algorithm
- a battle of forces
this document may be changed and modified at any time.
. . . . .
INTRODUCTION
- AI systems today utilises many different ML algorithms. These algorithms have proven good at finding patterns in data, yet what they lack is the dynamics of the thought. This algorithm produces the dynamics.
- The premiss for this project is that we accept that thoughts have mass (or an index from which we can deduce mass, force etc..). We talk daily about heavy thoughts.
- The base asumption is that we have access to an infinite number of thoughts, aka UNITS. From these we select an upper bound, which are then available for use in the System.
- The premiss for this project is that we accept that thoughts have mass (or an index from which we can deduce mass, force etc..). We talk daily about heavy thoughts.
- The base asumption is that we have access to an infinite number of thoughts, aka UNITS. From these we select an upper bound, which are then available for use in the System.
PROJECT VISION
- a new approach to AI.
- many other applications than ChatGPT.
- if accepted this may be a new paradigm in AI (and psychology).
- many other applications than ChatGPT.
- if accepted this may be a new paradigm in AI (and psychology).
PURPOSE
- to have my idea or vision confirmed and the generel setup validated.
- share ideas and new approaches to AI.
- have new inspirations for the direction of this project.
- get advise on how to optimize and better the solution.
- share ideas and new approaches to AI.
- have new inspirations for the direction of this project.
- get advise on how to optimize and better the solution.
THE VISION
- It all started with me recovering a trauma, which I noticed I kept going back to in my mind. This gave me the idea that thoughts could have a mass that was pulling me towards the heavy thoughts. I then scribbled down my idea, and it has become what it is today.
- One can imagine a black hole pulling in surrounding stars and planets. Since I dont have much expertice in quantum mechanics, I used Newtons equations and an analogy for for this system/situation instead which also produces a dynamics.
- One can imagine a black hole pulling in surrounding stars and planets. Since I dont have much expertice in quantum mechanics, I used Newtons equations and an analogy for for this system/situation instead which also produces a dynamics.
MECHANICS
Mech One:
- this is an analogy for the blackhole system. 2 cars connected by a rope or chain, they are going opposite directions and car one is dragging with a constant force, while car two is dragging with a variable force, which is dependant on the pull from car one. Here we can talk about "left or right" or "up and down". This produces a dynamic, which is used in this project for making and sending promts to ChatGPT.
Mech Two:
- Here one can imagine a ball balancing on top of a hill. The ball can go down the sides. The hill can be more or less steep. By pushing the ball up the hill, the game is to keep it from falling down the sides. Light thoughts are at the top of the hill and heavy thoughts are at the bottom. Like Mech One this produces a dynamic used to send promts to ChatGPT.
The similarities for these two mechanics, are that there is a static force dragging one direction and a variable force dragging or pushing the other way. ie. the car with a constant pressure on the the pedal (Mech One) and gravity (Mech Two).
From these two machanics we get a momentum, and if the momentum is above 0.0, the thought goes up and below the thought goes down. That is what we call THE CHOISE in this setup.
In later versions of the system, one can imagine entire landscapes of hills and valleys.
- this is an analogy for the blackhole system. 2 cars connected by a rope or chain, they are going opposite directions and car one is dragging with a constant force, while car two is dragging with a variable force, which is dependant on the pull from car one. Here we can talk about "left or right" or "up and down". This produces a dynamic, which is used in this project for making and sending promts to ChatGPT.
Mech Two:
- Here one can imagine a ball balancing on top of a hill. The ball can go down the sides. The hill can be more or less steep. By pushing the ball up the hill, the game is to keep it from falling down the sides. Light thoughts are at the top of the hill and heavy thoughts are at the bottom. Like Mech One this produces a dynamic used to send promts to ChatGPT.
The similarities for these two mechanics, are that there is a static force dragging one direction and a variable force dragging or pushing the other way. ie. the car with a constant pressure on the the pedal (Mech One) and gravity (Mech Two).
From these two machanics we get a momentum, and if the momentum is above 0.0, the thought goes up and below the thought goes down. That is what we call THE CHOISE in this setup.
In later versions of the system, one can imagine entire landscapes of hills and valleys.
FRICTION
- Since the static part of the mechanincs, always has the highest force UNIT, we multiply the calculated speed by a static friction (0.666) and on the dynamic part we multiply with a dynamic friction. Otherwise the static would always win.
- In order to calculate the dynamic friction, we use credits to find the friction coeficient.
- Now we can calculate the actual force for the dynamic part of the mechanics.
- In order to calculate the dynamic friction, we use credits to find the friction coeficient.
- Now we can calculate the actual force for the dynamic part of the mechanics.
FILTERS
the two main filters are:
Credit
- this filters task is to make sure not the same UNIT get chosen every time. Credit is continuesly updated. In order for a UNIT to be valid, it must have above 1.0 in credit. Every time a UNIT is "current UNIT" its credit reduces(fast) and is filled up again when not "current UNIT" (slow). Much like an infinity machine.
LowCut
- since the system had a tendency to pick heavy UNITs, I created this filter which sorts out the heaviest UNITs. This can be used to hide certain UNITs. Hidden UNITs are not unconciouss thoughts, but rather possible thoughts that have been lowcutted and therefore currently not available to the system. My vision was such a thought.
Credit
- this filters task is to make sure not the same UNIT get chosen every time. Credit is continuesly updated. In order for a UNIT to be valid, it must have above 1.0 in credit. Every time a UNIT is "current UNIT" its credit reduces(fast) and is filled up again when not "current UNIT" (slow). Much like an infinity machine.
LowCut
- since the system had a tendency to pick heavy UNITs, I created this filter which sorts out the heaviest UNITs. This can be used to hide certain UNITs. Hidden UNITs are not unconciouss thoughts, but rather possible thoughts that have been lowcutted and therefore currently not available to the system. My vision was such a thought.
ALGORITHM
- It is build on the notion: 1000(500) impressions, produces one thought.
- The system runs 500 times:
- - runs mechanics + friction
- - runs filters (Direction, LowCut, Credit...)
- - finds current UNIT
- After 500 runs, it finds the statistically most occuring UNIT. Which is then considered "the actual thought"
- The system runs 500 times:
- - runs mechanics + friction
- - runs filters (Direction, LowCut, Credit...)
- - finds current UNIT
- After 500 runs, it finds the statistically most occuring UNIT. Which is then considered "the actual thought"
CHATGPT
- the way I'm sending promts to ChatGPT, is that we play a game of "Connect The Dots". I give it 2 sentence (also created by ChatGPT, according to index and HUB) and ask it to connect the sentences, this produces the flow in the monologue.
WHAT IS IT
- At its core one can say it is a decision engine. Does it go up or does it go down and from this we can build more advanced systems.
- Other AI algirithms try to map every neuron in the brain, this algorithm instead maps entire networks(HUBs and UNITs) in the brain and use dynamics to move between these networks.
- Other AI algirithms try to map every neuron in the brain, this algorithm instead maps entire networks(HUBs and UNITs) in the brain and use dynamics to move between these networks.
EXPECTATIONS AND REALITY
- The algorithm is designed to handle an infinite number of thoughts, aka UNITS, I use about a hundred, so about ten HUBs. This makes the monologue a bit detached, but with more UNITS/HUBs it should seem more coherent.
- Further by grouping the HUBs better, it should give a more coherent monologue. Now for instance "programming" and "friends" are grouped together, which makes some awkward monologues. More closely related HUBs should give a more smooth monologue.
- This is still a prototype and the up down motion is a limited way of expressing the thought, and some other approach might be better suited. But i believe that modeling the thought as a dynamic is a valid approach.
- Further by grouping the HUBs better, it should give a more coherent monologue. Now for instance "programming" and "friends" are grouped together, which makes some awkward monologues. More closely related HUBs should give a more smooth monologue.
- This is still a prototype and the up down motion is a limited way of expressing the thought, and some other approach might be better suited. But i believe that modeling the thought as a dynamic is a valid approach.
LIMITATIONS
- no consiousness.
- no memory.
- no emotion.
- no free will (but the illusion of free will).
- explanation:
- no consciousness, others will have to figure that out.
- no memory and no emotion, this is purely the dynamics of the thought.
- no free will, due to the fact that the heaviest UNITs will be LowCutted, the system therefore cannot "see" the pattern or algorithm. Hence the illusion of free will.
- no memory.
- no emotion.
- no free will (but the illusion of free will).
- explanation:
- no consciousness, others will have to figure that out.
- no memory and no emotion, this is purely the dynamics of the thought.
- no free will, due to the fact that the heaviest UNITs will be LowCutted, the system therefore cannot "see" the pattern or algorithm. Hence the illusion of free will.
FUTURE OF THE PROJECT
- handle input from users, change subject, finish with a question, reject conversation, begin or lead the conversation.
- (Done)have ChatGPT dynamically create data for a UNIT, given a index. Thereby eliminating the 100 UNITs restriction and upgrading the system to fluent speech, but still only 10 HUBs (subjects).
- (Done)have ChatGPT dynamically create data for a UNIT, given a index. Thereby eliminating the 100 UNITs restriction and upgrading the system to fluent speech, but still only 10 HUBs (subjects).
MENTIONS
- as position goes towards 0.0 pain goes towards infinity
- the system is able to produce a random number, from momentum.
- this is my subjective vision of how the dynamics of the mind should be modelled.
- this is a prototype.. and therefore not the final version.
- the system is able to produce a random number, from momentum.
- this is my subjective vision of how the dynamics of the mind should be modelled.
- this is a prototype.. and therefore not the final version.