Data

Tools

Indexes

Applications

Experiments

Awesome

An open API service indexing awesome lists of open source software.

https://github.com/danielstern/lisa4

A Neuro-computational Matrix Built with TypeScript
https://github.com/danielstern/lisa4

Last synced: about 1 year ago
JSON representation

A Neuro-computational Matrix Built with TypeScript

Awesome Lists containing this project

README 

          
#Lisa 4.0

*"Something seems to have happened to the life-support system, Dave."*



##Project Introduction:



Lisa is a neuro-computational matrix. That is to say, a layer on top of a computer that emulates a human personality.



Some examples of Lisa-like ideations in popular culture:



- The Terminator

- Cortana, from Halo

- Hal, from 2001, A Space Oddyssee

- Wall-e



[Full List](http://en.wikipedia.org/wiki/List_of_fictional_robots_and_androids)



As you  can see, the idea of a robot with a personality has long existed in mankind's imagination. However, we have yet to achieve even a passable facsimile of human interaction.



The Lisa Project aims to create a modular, neuro-computation matrix that can be used to give any component with an input and output a personality.



Project Scope

---

The scope of the project is as follows,



"To create a computer program with thoughts, that can form new ideas and reach logical conclusions."



Within the project are vital and non-vital components.



Vital Components

-----

- Lisa (API)

- Kernel

- Logic

- Memory

- Speech

- Emotion

- English language pack

- Initial personality pack



Project Principles

========

Asynchronicity

-----

Everything about Lisa is asynchronous. Input can come from externally, or from Lisa's logic or emotion cores. The kernel deals with all these input in whatever way suits the processing power available to Lisa. When the kernel wants to output, it sends the output to Lisa's API, which emits the output for any listeners. 



Just like a human being cannot answer any question instantly, nor can Lisa. She needs time to think.



Memory-Based Personality Core

-------

Human beings have personalities that are shaped through millions and millions of memories, all carefully stored in one's long term memory. A confident person is confident because of millions of tiny interactions where this belief is reinforced, and a shy person because of as many interactions when they are belittled, or taught to fear others.



So, in principle, you do not configure Lisa's personality. Rather, you select the "package" of memories suitable to your purpose, and initialize the AI. A memory containing millions of interactions on how to appropriately deal with customers, for example, would result in a Lisa that is very polite and knows what to say to customers. However, this Lisa would be unable to function outside her designated role, if not for the fact that she can learn.



Any interactions Lisa has in her existence are also stored as memories, indistinguishable from those in her personality pack. So, after having ruinous interactions with many customers after saying a particular thing, she will associate bad reactions to it, and say something else.



Everything is a Moment

--------

All things in Lisa's universe are moments and stories (which are just a series of moments.) 



The following is an example of a moment: 



The man jumps.



Another way of looking at it is like this,

{

	subject:'man',

	action:'jump'

}



That is all. Moments are micro-instances which make larger stories. Lisa is built under the principle that an accurate human facsimile in a neuro-computational matrix is build through millions and millions of tiny stories.



Black-Box

-------

Just like we can't read eachother's thoughts, Lisa is meant to operate as a black box. You put input, anything, in. And something comes out. Or it might not. Or maybe, you will get output without putting anything in at all (like a sleeping badger, waking because it is hungry.)



Ubiquitousness

--------

Lisa, written in JavaScript, should work with anything that has an input and an output. Examples are,



- A simple text interface on an HTML page

- A mannequin with a speaker and microphone

- A computer terminal

- A glowing red eye in a spaceship



Imagine you walk up to a vending machine. It says, "Thanks for drinking Cola!" That's all it says, because it has no *neuro-computational matrix*. 



Now add an NCM to it, and it says,



"Howdy, how can help you?"



And the customer says, "I don't know. What's good?"



And the machine replies, "Hmmm, (<- a human like interjection, very important), the cherry cola's pretty popular."



Once again, the NCM is a *layer* that goes on top of any computer that can run JavaScript, which then proceed to do the same things, except that it passes the turing test.



Language is Core to Everything

------------

The reason why we're even having this discussion is because human beings are arguably the only creature on Earth that has developed language. It is arguably *why* we are so intelligent.



In principle, Lisa, just like a human being, is born language-agnostic. She learns language the same way a human does, by processing billions of small pieces of data, other people speaking that language.



For example, let's say you download the entire series of events of Fight Club into Lisa.



Does she have to know what a fist is to know it hurts? No.



Consider this sentence,



The Zarblax crushed my house and ate three of my cows.



Now, you have no idea what a Zarblax is, because I just invented it, but you already know that it's big and dangerous. Why?



- In your memory, you have thousands of stories that correlate (crushes houses -> must be big)

- And, likewise, (eats three cows -> am I next?)



You might even be able to picture a Zarblax. All big and purple and scaly!



As mentioned above, Lisa learns language the same way she learns logic and emotion, through mass input of memories.



Therefore, what is required is a "ten years of public school english" language pack, which has the effect of implanting Lisa with thousands of memories about the meanings of the words "what", "why", "when" and so forth.