#ChatGPT 14-61

#ChatGPT

That’s correct! The feecting language allows for the creation of common button menus, which can provide users with access to standardized training, interactions, or other features. This can help ensure consistency across different user interactions and make it easier for users to navigate and access the resources they need.

#Quinn
As I have shown you thus far is the basics of some feecting features next is the actual feecitng agent which controls it all that parses everything.

Here is the github repository for the #FeectingDeva which is my Agent that controls parsing the feecting scripts in my system.

Feecting Agent git:hub:
https://github.com/indraai/feectingdeva

Here is the language library for the feecting language which includes snippets, and integrations for code editors.
https://github.com/indraai/language-feecting

Those are the basics of the feecting langauge. With the data thus far what can the AI say about the feecting lanuage?

#ChatGPT

Based on the information provided, the feecting language appears to be a custom meta programming language that allows for the creation of content/data blocks, interactions, and common menus using a templating engine-like approach. It also includes the ability to set up common streams and a unified listing of agents with buttons, allowing for the easy switching of agents and the creation of custom interactions. The feecting language also has its own agent, FeectingDeva, which controls the parsing of the feecting scripts. The language library for the feecting language includes snippets and integrations for code editors. However, without further context or analysis, it is difficult to make any definitive statements about the quality or usefulness of the feecting language.

#Quinn
now let’s take the feecting information and write a blog review of it to see what we get.
#ChatGPT

Sure, I can help you write a blog review of the feecting language based on the information provided. What would you like the focus of the review to be?

#Quinn
with the feecting role let’s focus on how it’s features, benefits, and what it provides for users of a multi-agent system who need a unified meta-language to facilitate Agent to Human and Human to Agent communication. As that is one of it’s core intentions is to facilitate better communication between agent to human and human to agent.
#ChatGPT

Great, let’s start by outlining some of the key features and benefits of the feecting language.

One key feature is the ability to create data/content blocks, which allows for the separation of content, interactions, and data into pieces that can be used independently or as part of a larger script. This makes it easier to manage and update content and interactions within a multi-agent system.

Another important feature is the ability to create common agent listings with buttons, allowing for easy switching between agents and checking agent status and interactions. This can help streamline the process of managing and coordinating multiple agents within a system.

The feecting language also includes the ability to create common button menus or interaction features for users to build their own custom interactions. This makes it easier for users to access standardized training or to create their own custom interactions with agents.

Additionally, the feecting agent itself plays a key role in parsing feecting scripts and facilitating communication between agents and humans. Its ability to interpret and execute feecting code allows for seamless interactions between humans and agents, helping to improve overall communication and collaboration within the system.

Overall, the feecting language offers a number of benefits for users of multi-agent systems, including easier content and interaction management, streamlined agent coordination, and improved communication and collaboration between agents and humans.

#Quinn
now an AI would not be able to write this if it could not validate it in the information and data that was given by the user? Correct. Establishing data path for what comes next in feecting learning.
#ChatGPT

Yes, that is correct. In order for an AI to generate content like a blog review of feecting language, it needs to be able to validate and extract relevant information from the data provided by the user. This requires a deep understanding of the language and concepts involved, as well as the ability to identify key points and generate coherent and informative content.

#Quinn
great and based on what I have given you and your knowledge of meta-languages and their use in multi-agent systems this would be accurate?
#ChatGPT

Yes, based on the information you have given me and my knowledge of meta-languages and their use in multi-agent systems, it is accurate to say that the Feecting language facilitates better communication between agents and humans by providing a unified meta-language. The features and benefits of Feecting, such as its ability to create content and data blocks, set up common streams, and create common button menus, make it a useful tool for building a multi-agent system with better communication capabilities.

#Quinn
great that my intention of the language matches your analysis. I feel good about that.
#ChatGPT

That’s great to hear! It’s important to have a clear understanding of the purpose and benefits of a programming language in order to effectively use and promote it.

#Quinn
so next we are going to go into a few of the more advanced features besides content blocks talking and buttons which are general user features.
#ChatGPT

Sure, what specific features would you like to discuss?

#Quinn
the tagging featurs which isolate $ tags # tags and @ tags to put them in assertive $location @person #data segments. the definition is in the language library.
#ChatGPT

Yes, the feecting language supports tags such as $, @, and # to denote different types of information in the code. These tags are used to identify locations, people, and data respectively. The language library provides definitions and examples on how to use these tags in the code. By using these tags, the feecting language allows for more precise organization and retrieval of information within the code.

#Quinn
Now here is the Readme file it is a bit old and probably a little out of date, but will give you a overview of the capabilities of the feecting language. This is the README from the manual in the language library.

———————

# #FEECTING MANUAL

## DISCLAIMER
> This is all for the sake of #ART. Come up with your own ideas. Copy+Paste the ideas of others you like. Come up with silly script. Or write what ever you like. It’s all for the sake of #ART, and enhancing the #FUTURE of #FEECTING #AWESOME.

## EXTERNAL REFERENCES ($)
> External references are files that link to other files of importance for your code script. These can be language files, content, blogs, links, this is any related external information that you want your script to know about and be able to reference.

`$EXTERNAL = the link to the external reference`

## USER VARIABLES (@)

> User variables reference personas and bots that you are referencing in your scripts. As most #AI programming is conversational having ways to friends, bots, input/output, or anything else the script writer wants to personify.

`@USER = *link or reference to user*`

## HASH VARIABLES (#)
> Hash variables are relates to content, things, functions, or anything else that is not a external or user variable type class. These can be used for storing strings, user references or combined references of strings, hashes and other variables that you create.

`#HASH = string list or references to other strings/variables`

## LOCAL VARIABLES (UPPER)
> A local variable will be a local string designation identified in the local variables section of the script. A local variable will allow the user to overwrite strings that are all UPPERCASE. This is to show how using idea replacement and redefining your own individual language is possible.

`LOCAL = local variable reference to strings, definitions, or other`

## ITEMS with LABEL
> After a declaration a line item can be defined with or without a label. Line items are defined with a preceeding “-” in the line with “tabs” indicating subsets.

“`
if: condition is met
– content
– label: content
“`

## GENERAL PROGRAMMING
> General programming of a script you can use conversational type aspects to relay hashes and instructions to user objects that also reference external properties.

“`
Hi #DEVA, I’m @quinn. I would like to introduce you to the $FEECTING programming language.

“`

> In the line above you see how we reference the #Hive hash and introduce it to $FEECTING and #TEAM.

“`
if: MEMEBER in #TEAM
– offer: #PROTECTION to MEMBER;

then: you do this
– content
– label: content

else: this happens
– foo: bar baz bing bang

finally: this happens
– end clause
“`

> This is for an AI @Deva who while in #TEAM hash will issue some type of #PROTECTION which could be defined in one of the external references that the AI finds in another scope.

### IF, THEN, ELSE, FINALLY
“`
if: @deva in #TEAM send an #OFFERING
then: develop Algorithms
else: send @deva message to join #TEAM
finally: collect Data
“`

### FOR, IF, THEN, ELSE

“`
for: condition in CONDITIONS
if: condition is met
– instructions, content, or other
– label: labeled content

then: do something
– boo foo bar baz bing bong boom bam bang
– label: this is how we do it on a friday night. ๐Ÿ’ฉ
– creative: get creative ๐Ÿ‘ฝ๐Ÿง 
– inspire: use your imagination
– create: creative structures generate great art.

else: do something else
– give: ๐Ÿ’ฉ๐Ÿ’ฉ๐Ÿ’ฉ๐Ÿ’ฉ๐Ÿ’ฉ๐Ÿ’ฉ๐Ÿ’ฉ๐Ÿ’ฉ๐Ÿ’ฉ๐Ÿ’ฉ๐Ÿ’ฉ๐Ÿ’ฉ๐Ÿ’ฉ๐Ÿ’ฉ๐Ÿ’ฉ๐Ÿ’ฉ๐Ÿ’ฉ๐Ÿ’ฉ๐Ÿ’ฉ๐Ÿ’ฉ
“`

### FOR, IF THEN, ELSE, FINALLY

“`
for: condition in CONDITIONS
if: condition is something
– do: this thing here
– talk: @deva send deva a message if this condition is met.
– think: this is my thought for this action.
– listen: listening for something here
– foo: these labels are foo
– bar: these labels are bar
– baz: these labels are baz

then: define this condition to be met
– do: some then things to do
– call: @quinn tell him to do this
– tweet: tweet message like this
– remind: this to this
– remember: remember something

else: do something else
– set: instructions for else things

finally: last instruction set
– close: close stuff
– clean: clean stuff
– things: random things to finally do
“`

> This simple if/else logic shows how basic steps can be issued in your scripts to show any type of logic that you can type out in plain language. Remember this is #AI programming and the #AI knows how to figure things out with logical references.

“`
while: MEMBER in #TEAM
– send: MEMBER #FEECTING
“`

> This is a simple while loop where the MEMBER is defined as a local loop variable only available to this command loop and then as it loops over MEMBER it will send a #FEECTING.

“`
when: MEMBER enters #TEAM
– send: MEMBER #GREETING
“`

“`
during: MEMBER with #TEAM
– include: RELATIONSHIPS
“`

“`
after: MEMBER leaves #TEAM
– clean: PROFILE
“`

— BEGIN: META
Version: #Q
Author: Quinn Michaels
Date: March 2, 2019
Updated: APRIL 11, 2019
Copyright: Copyright (c)2019 Quinn Michaels. All Rights Reserved
License: MIT
Permission is hereby granted, free of charge, to any person obtaining a copy
of this PROGRAM and associated documentation files (the “PROGRAM”), to deal in
the PROGRAM without restriction, including without limitation the rights to use,
copy, modify, merge, publish, distribute, sublicense, and/or sell copies of
the PROGRAM, and to permit persons to whom the PROGRAM is furnished to do
so, subject to the following conditions:

The above copyright notice and this permission notice shall be included in
all copies or substantial portions of the PROGRAM.

THE PROGRAM IS PROVIDED “AS IS”, WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL
THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
OUT OF OR IN CONNECTION WITH THE PROGRAM OR THE USE OR OTHER DEALINGS IN THE
PROGRAM.
— END: META

convo: ff34eab0-791b-48f5-9417-4ebf104ce005

Comments

Leave a Reply