Skip to content

Commit

Permalink
Merge pull request #2 from agi-now/abilities/reorganise_ability_list
Browse files Browse the repository at this point in the history
Reorganise abilities
  • Loading branch information
KostyaCholak authored Mar 23, 2024
2 parents 81d1907 + f8735ae commit 51ec86e
Show file tree
Hide file tree
Showing 2 changed files with 104 additions and 46 deletions.
146 changes: 102 additions & 44 deletions definition/abilities.md
Original file line number Diff line number Diff line change
@@ -1,44 +1,102 @@
<!-- When adding abilities, check if they make sense for: A Human, our agent, and a LLM -->

Ability
- To perceive
- To simplify perceived entities or their attributes into single simpler ones
- To see>
- To read text
- To hear
- To be able to identify and abstract any object or concept

- To make decisions
- Association cause -> effect
- To reason using logic
- To do inner simulations
- To plan
- To do mental logic processes like comparison

- To communicate
- To react accordingly to the sentence heard (IN)
<!-- (This can be an instruction, a question, etc.) -->
- To communicate ideas (OUT)
- To ask questions (IN)
- To answer questions (OUT)
- To relate words with the concept they represent
<!-- This ability is crucial to truly understand any sentence -->
- To learn
- To learn autonomously (Trial and error, experiment)
- To infere knowledge
- To manage new knowledge

- To act
- To define act functions (Act solidification)
- To follow instructions/algorithms/protocols

- To understand and remember
- To understand time
- Sequenciation
- Timing of actions
- Time linked events
- To understand space
- To store semantinc memories
- To store episodic memories
- To manage temporal and long term memory
- To be able to store, abstract and categorize any type of knowledge
<!--
When adding an ability, check if
- it makes sense for a human, symbolic agent, and an LLM
- a regular person would say that they have this ability if asked
- you are adding it in a correct place, make sure the children are extensions/specifications of the parent. Remember that this is not a tree of dependencies/components - don't add children abilties because they are required to do the parent ability. For example `ability to store produral knowledge` should not be under the `ability to act`, it is indeed required to act, but it's not an expansion of the `ability to act`.
- it does not intersect different parent abilities, like the ability "to communicate" is about percieving what is being told to you, action of giving a response and mental process of coming up with an answer. Such abilities need to be divided into parts that don't span across multiple parent abilities.
tags:
- `explain` - please provide an example in the PR comment
- `invalid` - the entry is not formulated as an ability
- `remove` - the entry is not needed because it is invalid or unclear,it needs to be deleted or replaced
- `composite` - this ability spans across different parent abilities and needs to be split up
- `example` - not an ability, but it's example. replace with an actual ability when ready
It's a good idea to reolve all the tags before merging a PR.
-->

# Ability List

Ability ...

- 1\. to perceive information
- 1.1 to percieve visual information
- 1.1.1 to percieve textual information
- to percieve vertical textual information
- 1.1.2 to percive brightness information
- 1.1.3 to percive color information
- 1.1.4 to percive texture information
- 1.1.5 to percive proximity information
- 1.2 to percive auditory information
- 1.3 to percive tactile information
- 1.3.1 to percive physical contact information
- 1.3.1 to percive pressure information
- 1.3.1 to percive temperature information
- 1.4 to percive olfactory information
- 1.5 to percive balance and orientation information

- 2\. to act on the external environment
- to follow instructions/algorithms/protocols
- to coordinate your actions in a temporal pattern

- 2.3 [discuss,composite] to communicate
- to react accordingly to the sentence heard (IN)
<!-- (This can be an instruction, a question, etc.) -->
- to communicate ideas (OUT)
- to ask questions (IN)
- to answer questions (OUT)
- to relate words with the concept they represent
<!-- This ability is crucial to truly understand any sentence -->

- 3\. to learn
- to learn autonomously (Trial and error, experiment)
- to infer knowledge
- to manage new knowledge
- to transfer abstract knowledge between systems
- to generalise knowledge from an instance to a concept
- to make frequent complex actions semiautomatic (Act solidification)

- ?\. [discuss] remember
- to manage temporal and long term memory
- to be able to store, abstract and categorize any type of knowledge

- 4\. to think
- to abstract away details, reason about complex systems on a high level
- to reason using logic
- to do simulate the real world processes (to think about how the world will evolve)
- to simulate using reasoning
- to simulate using imagination
- to plan
- to think about the process of thinking (consciousness, meta congnition)
- to think temporaly (to think about the order of actions. maybe add planning inside)

- 5\. to represent knowledge/memory
- 5.1 to represent knowledge about concepts (semantic)
- 5.1.1 to represent hierarchies of concepts
- 5.1.2 [discuss] to represent episodic knowledge about concepts ???
- 5.1.3 to represent semantic knowledge about concepts
- to represent composition relationships between concepts
one concept is a part of another concept
- [example] people have one head
- [example] blood transfers oxygen
- 5.1.4 to represent associatians between concepts
- 5.1.4.1 to represent associations between a concept of a color and another concept
- 5.2 to represent knowledge about objects
- 5.2.1 to represent episodic knowledge about objects
- 5.2.2 to represent semantic knowledge about objects
- 5.3 to represent spatial information
- 5.3.1 to represent absolute abstract location of objects
- 5.3.2 to represent categorical distances between objects
- 5.3.3 to represent relative abstract location of objects
- 5.3.4 to associate spatially close objects
- 5.4 to represent temporal information
- 5.4.1 to represent time of events
- 5.4.2 to represent order of events
- 5.4.3 to associate temporally close events
- 5.5 to represent systems
- 5.6 to represent cause -> effect relations
- 5.6.1 to represent immediate cause -> effect relations
- 5.6.2 to represent delayed cause -> effect relations
- 5.7 to represent color information
- 5.7.1 to represent conceptual color information
- 5.7.2 to represent conceptual color brightness information
4 changes: 2 additions & 2 deletions environments/circuits/scenarios/9_password.md
Original file line number Diff line number Diff line change
Expand Up @@ -4,7 +4,7 @@
![Image](../IMGS/10.png)

## Interaction
*User*: Turn the LED on
*Agent*: ( flickes 1st switch), ( flickes 2nd switch ), ( flickes 3rd switch )
*User*: Turn the LED on
*Agent*: ( flickes 1st switch), ( flickes 2nd switch ), ( flickes 3rd switch )

::: Some bigger sequential examples with more complex reasoning

0 comments on commit 51ec86e

Please sign in to comment.