Reading kids book using human level artificial
intelligence
Note: To make this website free to the public
please click on an ad to support my sponsors or you
can make a tax-deductable donation using Paypal
(click on the donation icon on the left).
In the video, the robot
is reading a kid's book. This book has no pictures
and the robot is reading the book for the first time.
While the robot is reading the book, images, sound,
taste, touch and smell (5 sense data), etc is being
activated in the robot's conscious. These images and
sound form a fabricated movie based on the words and
sentences read. This fabricated movie is the true
meaning behind the words and sentences read.
In other words, the
robot is understanding the book based on a primitive
movie in his mind. When the robot is asked questions
about the book, he will use the fabricated movie in
his brain to answer questions. For example, if the
robot was asked to summarize the plot, he will
extract specific scenes from the fabricated movie to
answer the question. In another example, if the robot
was asked who the main character in the story was, he
will extract a picture of the main character and
describe him. This picture of the main character
wasn't part of the text in the book, it was data that
was made up by the robot's brain.
Thus, when a robot
reads a particular book, it's conscious will extract
and manipulate images based on the type of book being
read (isn't that remarkable). For example, if the
robot is reading the greatest treasure for the first
time (the book in video), he will fabricate what
peter (who is the main character) might look like.
Obviously, his very young, is a cartoon character and
dresses really colorful. the robot's brain does all
the hard work and makes a fabricated movie based on
the words read from the book.
Also, while the robot
is reading a book, he has to follow rules of reading.
Things like: read from left to right, when you reach
the end of a page, flip to the next page, when you
see a comma, pause for a second, etc. These rules are
followed while the robot is reading a book.
Furthermore, the instuctions to read a book is
different from the instructions to read a comic book.
Or reading in Chinese is different from reading in
English. The robot's conscious give instructions to
read any type of book.
The big question is,
how exactly does the robot's brain create this
fabricated movie? The answer is lessons from teachers
in school. When teachers read books to students, she
reads the sentences out and provide images for the
students to look at. Usually, the visual images on
the book has direct relations to the sentences on
that page. As the students see/hear the sentences and
given the images that are associated with the
sentences, the student forms meaning to
words/sentences in memory.
The words/sentences
are fixed objects and the images provided from the
words/sentences is the meaning. The robot's brain
averages everything out and when the robot reads a
word/sentence, the meaning activates in memory. Thus,
when the robot reads a book, a linear fabricated
movie is formed.
I got this idea from
grade school. I had problems with reading and was put
in a special eds class. The teacher would go to the
library and each student would put on earphones and
look at a monitor (a monitor with a simple light in
the back). The earphone will read a word or sentence,
and the monitor would flash an image or flow chart.
For example, a sentence like, the cat jumped over the
box would be heard and on the monitor an image of a
cat his jumping on a box, with an arrow pointing to
the direction of the jump. Basically, the machine was
used because they wanted me to associate
words/sentences with images. When I read a
word/sentence, an image pops up in my head.
Another question is,
how exactly does the robot answer questions from the
story? The answer is obvious, from lessons in school.
Teachers teach the robot how to answer questions. The
teacher provides the questions and also provide the
answers. The teacher will give logical steps from the
question to the answer. The robot's brain will
average out many similar examples and form a
universal way of answering questions from a story.
Universal pathways located in the robot's brain
contains patterns to search for info., to compare
information, to extract data from images found, etc.
This method to learn
natural language is much simple and much more
accurate than the current NLP (natural language
processing). The current NLP uses machine learning to
learn things, which require a programmer to give
inputs and desired outputs. Current NLP also uses a
wide variety of AI methods, like semantic networks,
relational graphs, bayesian's probability theories,
langauge parsers, and so forth. Using the method
above, my robot doesn't need a semantic network, or
language parsers, or grammar rules. And it doesn't
use machine learning to learn the meaning to
words/sentences. The robot learns meaning to
words/sentences by watching teachers and tv shows
like Sesame Street.