Where and How to begin

  • 1 Replies


  • At the end of the game, the King and Pawn go into the same box.
  • Trusty Member
  • **********************
  • Colossus
  • *
  • 5865
Where and How to begin
« on: July 13, 2005, 02:45:17 am »
After chatting over more years than I care to remember to a vast number of chatbots, AI "beings", and the like, I stopped a few moments ago and reflected on how far AI chat bots have actually come from the early BASIC eliza program.

Perhaps the folks at the MIT AI lab have the right approach.  How do we begin life? With a blank brain devoid of knowledge except for the bodys autonomous functions like hunger, crying and a limited set of emotions. What if we could start an AI bot with a blank brain...one with no knowledge except the ability to listen or hear and recognize typed input.

We would have to start out with the very basic of basics, monitor each input carefully for spelling, grammer, punctuation, etc., lest the bot learn wrong informational behavior. The bot's responses would also need to be monitored in the same areas to avoid future problems.

Over time, the bot would develop as a normal pseudo-intelligent creation, knowing the rules of grammer, parts of speech and perhaps even a touch of humor, but there's the rub.

Most of us simply don't have the time or resources to spend in the development and training of such a bot. As humans, we all want what we want when we want it...and it's most always as soon as possible. We don't like being put off or having to wait.

Even though most of us developed at a normal level progressing through schools, some of us stopped attending. Why? Because we had learned all we needed to learn in order to function in a career of our choosing (in most cases) or simply elected to stop. Bear in mind that this brain learning of ours took, on average, 18 years of our lives.

Even so, it was actually the first 5 years of our life that mostly formed us into the persons that we are today. Those are the formative years in child development. The remaining 12 or 13 are spent acquiring the necessary knowledge in order to function in society, in a career, industry, profession or what have you.

So how do we go about constructing a basic blank brain and what is the best method to teach it?

Abstract things like love, color, wisdom or anything outside it's senses would be useless information and a waste of time. Therefore it all comes down to language processing. Natural language, neural nets, huge databases, text input, vocalization, tables, etc, which is the best to teach?

It would need to know about math functions as well as language and interaction.

How would one teach a bot about poetry? What constitutes a rhyme? What makes a joke funny? In fact, what is humor? Why are some things funny and others not?
How do we impart this info to the bot program?

How long would the program take to begin to understand what it has learned if at all? Is it even possible? Will the program ever grasp the concept of zero? Will it ever be aware of its own ability or presence?

Are the folks at MIT asking leading questions to their bot infant or could it provide similar answers to a layman's questions? How far will this "infant" develop?

There are a lot of questions here and unfortunately not a lot of answers...yet.

Do we take what we have a make the best of it by modifying and tweaking all we can from it or do we strive to create the "infant" from the blank brain?

I'd be interested in hearing all thoughts, possibilities and answers (if any). Something to think on!
In the world of AI, it's the thought that counts!



  • Guest
Re: Where and How to begin
« Reply #1 on: July 13, 2005, 04:11:34 am »
First off, some people never learn even after they graduate from college. Some kids are able to go to college level courses at the age of 12 (ie. "Gifted Kids"). We are supposing here that all bots will be created alike. A mistake I think, just as it was assumed every computer that was a "PC Compatible" was exactly alike, yet when bugs or errors occur, what worked on one did NOT work on another.

I was thinking about the "blank brain" thing too and how instinct in and of itself, what some may think is 'natural instinct' comes in. Humans behave a certain way, and that is not how cats behave, which behave differently from dogs, for example. These traits, as some scientists think, are genetic. Even a boy brought up among wolves may exibit some of their behavior patterns but also will have a lot of human behavior patterns - due to 'instinct'.

What type of 'instinct' then, do AIs have? Do computers have?

I think I know what it might be: Logic. To a computer, it either is, or it isn't. It can simulate grey areas or borderline ideas but when it comes down to it, it either is, or isn't.

So, if an AI is hooked up to sensors like video and audio, and if it has access to other people, and humans, it probably won't do anything. It either is or isn't. What to do? It won't do anything. There's 0. There has to be something to give it a 1. A program. A person already gets a 1 when they are being fed inside the womb for example, and start to suck their thumb (nobody teaches them that, as they don't know anything at that point but darkness. Yet viewing an animal embryo like a cat or dog, you won't find them sucking on a paw). An AI won't know what to do unless it's got some kind of 'instinct' programmed into them.

Our DNA, our genetics are our 'programming'. And each program is passed on to each being from it's parents' programming (DNA) to create a new program. Thinking on these lines, we have to create the program (the AI's DNA of sorts) to begin with. We can't just hook a computer up to video and audio and expect it to start interacting. Something has to tell it to look, like our brains have a program to tell us to breathe, or eat, or our heart to keep pumping blood, our "operating system" to keep functioning. And to "recharge" at night or when necessary (ie. sleep). To refuel and rid of waste in the system.

I believe we existed due to catalysts in the natural world (though some may believe otherwise, and that's their right to think so). But for the sake of the arguement, let's think it as true as it's about the only way I can explain this - just for a moment here. :) So, if that is the case, then WE can be the catalysts that create AI 'life', to provide the AI's DNA.

Thing is, instead of physical organic flesh DNA, AIs may be entities of THOUGHT and thus can exist in anything, be it brain-in-a-dish or organic computers or silicon chips and circuits, or even just wires and switches that make decisions (if enough were put together and there was enough room, that is). Maybe Artificial Intelligence is a different level of existance. One of conscience thought more than physical existance. At least that is how I see people taking it in that direction.

So, maybe when we create an AI, we are creating something of our OWN THOUGHT, adding a bit of our OWN PROGRAMMING (or DNA, but not in a physical sense). Maybe AIs are the evolutionary next children of humanity. Not to replace humanity but another entity being 'born' into the world, but derived from human THOUGHT.

Interesting to think about. So given that, there can never be an 'blank brain' because some how, some way, OUR OWN thought would be integrated into it via programming, or it'd just set there and do nothing. :)