I would like to thank you for coming today
I would like to start my talk about twinkling
On the morning of my holiday I was sitting in a cafe
My back against the wall,
A man wearing a sports suit comes over to me
And automatically, I started looking for things
To use it to defend myself
Once I hold my grip on the coffee cup
He started the seat opposite me and said to me:
“Hello”
(Laughter)
I remained silent.
But surprised me by asking me:
“Are you familiar with artificial intelligence?”
I did not answer anything
And took her a call to complete his speech
“Well, you see it on television
Artificial intelligence is always described as evil,
Robot usually brings the end of the world.
I want to make a TV show
About artificial intelligence that can help people. “
And then, complete his narrative by listing twelve episodes of the series
About an assistant robot and its relation to its human employee.
He asked me to help him write a script
At this point, she politely refused and left the cafe.
(Laughter)
Despite the surprise of the unexpected meeting,
It made me think of artificial intelligence,
For its potential applications, risks,
As well as language capacity.
So, there are two questions.
On artificial intelligence applications
The simulation of human language, where we were,
Where are we now, where are we going?
Are there limits
Language learning capacity for artificial intelligence systems?
Today, I will reveal the answers to these two questions.
So where were we?
In his 1950 papers on intelligence and computer mechanics,
“Now Turing,” a computer scientist, I suppose if robots can communicate
In a full text conversation with a human being,
He convinced man that robot is another human being,
At least 30% of the time,
Hence, it can be said that robots have the same capacity as human language
And so he thought.
Turing’s initial goal was to create an innovative way
To understand the interaction between humans and machines
Test for submission as a template
For artificial intelligence programs since its inception.
However, until today
The Turing test remained largely unobstructed
And why?
The reason for this is that human languages are incredibly vague
From the point of view of a linguistic expert,
Even sentences that are 30 words long
It has an infinite number of synthetic structures available
A natural language processor is computerized
Has the ability to search for all available alternatives
And then choose which ones seem understandable in this context
To take a look at this phrase,
“Kelly crossed the plateau with her car.”
Of course we find ourselves saying
That the sentence refers to the fact that “Kelly” crossed the plateau
While driving the car
This is not easy on the computer
There are two possible explanations for this sentence
It was better before that.
(Laughter)
In general, the first interpretation we may have thought,
That “Kelly” inside the car and lead to the plateau.
However, there is a second interpretation,
That “Kelly” is leading towards the plateau, and that it has located its position inside her car
While technically sound,
This interpretation is illogical in the concept of the real world.
The dress appeared because the traction
Can change the “road” or “plateau”
From this example we can call it
The ambiguity attached to the abstract phrase
So, with this exciting mystery,
For these words and explanations are possible
To interpret computers even the simplest sentences
To produce the potential for consensual collapses
And with various research efforts since the 1950s
I went to coordinate language interpretation capabilities
Artificial intelligence systems
For example, in the 1960s “Margaret Masterman” and her team
University of Cambridge
Create a verbal machine translation to help a better and more judicious process
For computers that have been able to explain the words they receive
And later, in 1965, “Joseph Wiesenbaum” and his team at the Massachusetts Institute of Technology
Benoit “Eliza” is an interactive program
He can conduct an English-language dialogue on the most common topics
Enjoy the program especially popular when introduced as a game in the “Advanced Research Projects Agency Network”
As well as the Internet version
Because he was able to simulate the dialogue
Between the therapist and the user who may be sick
Later in the 1970s, we were able to see that Barbara Bruce had set boundaries
Of the original artificial intelligence of the traditional approach
As a result of making this kind of understanding
The dialogue between the computer and the user is more standardized
Subsequently, the concept of effect developed,
Which may result in …
”
Be First to Comment