.jpg)
.jpg)
Welcome to the digital frontier – GPT3 allowing AI to be more human
By Andrey Demiyanov
- +
- -
AMMAN — In today’s ever evolving technological age, the term
artificial intelligence gets thrown around more than a soccer ball on a field
at football practice. With AI being all the rage, startups and mega
corporations continue to strive towards implementing human-like intelligence
into their everyday processes in order to increase efficiency, profitability
and scalability of their operations.اضافة اعلان
Artificial intelligence as a term was first coined by John McCarthy in 1956 where he held the first academic conference about the topic. While the concept of AI was previously discussed in theory years before McCarthy’s presentation, 1956 was the year that truly shifted the conversation from generalizing about the concept to fully evaluating its potential, threats and benefits to society.
In 2021, artificial intelligence is ever present, without you even noticing it. Amazon’s recommendations based on prior purchases, algorithmic trading and chat bots are some of the many implementations of AI in today’s society.
In this article we discuss GPT3, a neural network machine that has been trained via millions of data sets on the internet to read, comprehend, analyze and generate text based on linguistic input. The network was designed by OpenAI, and is by far one of the most sophisticated text-generating machines available to the world to date.
What is OpenAI?
OpenAI is an artificial intelligence research laboratory founded by several individuals, most notably Elon Musk. The purpose of the company has always remained to create and research artificial intelligence for the benefit of all humanity.
OpenAI has multiple projects that it has developed over its lifecycle. As a for-profit company, many of their projects are available for businesses to use through their website or through direct contact. That means businesses can use the latest innovations from one of the leading corporations on AI around the world at any given time.
One of their latest projects, OpenAI Codex, is an AI system that converts natural language inputs into code, which means that, hypothetically, it is capable of enhancing the speed at which developers generate code. It also means that less technologically adept coders would be able to generate elements or components to add to their code through basic linguistic input.
What is linguistic input?
Humans communicate through a variety of languages, whether Arabic, English or French, for example, or any of the hundreds of languages and dialects from around the world. Language as we know it is a purely human concept that we have adopted for the purposes of communicating with each other. Computers however speak a different, unified language: binary, where two digits define the underlying message of every sentence through ones and zeroes.
(Photo: Brent Walsh)
All those ones and zeroes represent values that are linked to each character within the text. Through these ones and zeroes, a computer is able to take inputs made by a human and understand them in order to perform an action.
In short, while you may speak one language, the computer is only ever able to comprehend binary. Here’s where it starts to get interesting.
Inferring meaning from language: a very human skillset
While most people understand jokes, metaphors and infer meaning behind text, a computer has a significantly harder time understanding chains of text altogether. When talking about a “grey chair” you would think of perhaps a grey colored chair, possibly situated in some kind of a room, probably in a shade of grey that others wouldn’t have thought of. The blanks that each of us fills in, the room, the shade of grey and maybe the shape of the chair, are all things that we infer and is second nature to humans. To a machine, however, this is no trivial task.
This is natural language processing (NLP) fits into the picture. NLP is the process of taking natural language that humans use and processing it in ways that would make it understandable for a machine. Unlike “dumb AI” of the past, which would look for specific key words or sets of words, an AI using NLP may infer meaning behind text and make assumptions about meaning, which is a critical component to language. However, as with all artificial intelligence, large amounts of data must be amassed in order to train the algorithm.
This is why the internet has played such a key role in developing artificial intelligence. All of our Google searches, all inputs we make on websites that collect our data, everything we do has been gathered, sorted and fed to algorithms. This I show a machine learns how to read, understand and speak “human”.
Early iterations of artificial intelligence were primarily based on mathematical functions that were sought to interpret specific data into specific results. Modern AI is capable of taking a more holistic approach to language and its meaning, interpreting and inferring how language is used and underlying undertones in order to generate responses.
In other words, modern AI machines not only understand what we’re saying, but infer the hidden meaning behind our words, and uses this knowledge to produce highly relevant results.
How does this affect me?
Machines are able to go through a large amount of data in seconds, analyzing, determining and inferring meaning behind the language used within a text. Hypothetically, this means that anything we may request from an AI that is sufficiently trained and is capable of performing tasks, can be carried out on a whim.
Having machines understand the human language in its full context has been a challenge since the inception of the concept of AI back in 1956. For all intents and purposes, the perfect AI acts, behaves and speaks like a regular human being, with the added benefit of being hooked up to a large pool of data available for processing, in this case the internet.
In essence, a truly powerful AI that is capable of mimicking human behavior and understanding, is something that modern day scientists strive for more than anything else due its potential in changing the overall technological landscape. General Intelligence, or GI, is what modern day AI researchers consider machines that are capable of constructing and displaying cognitive human-like abilities.
When prompted to “Write this as an Attorney” and given only the first two sentences as prompts, the GPT3 managed to write this masterpiece. (Photo: Francis Jervis)
Humanity is still far away from the GI technology needed to write this article; however, every day we move closer to the inevitable point where humanity and machines will be able not only to communicate but understand each other at a level that we assumed only humans were capable of.
What is GPT3 and why does it matter?
Before we go into GPT3, we must first understand what a neural network is. A neural network is a computing network that has been developed in tandem with biological neural networks found within animal brains. In other words, its essentially the brain’s infrastructure as we know it implemented into the digital market place. It is used to educate, train and evaluate a computer’s output paired with inputs in order to further its knowledge on subjects.
GPT3 is one of the most complex natural language processing models in the world that has a natural language generation component. This component is capable of not just understanding concepts, both implicitly stated and inferred through text, but also provides highly relevant and contextual output in text form as well. It’s one step closer to a truly human machine.
Aside from working in tandem with GPT Codex to create functional code, GPT3 is capable of generating language outputs akin to that of a real human. Through the use of over 175 billion machine learning parameters, GPT3 is far more advanced than its predecessor, the Turing NGL Model, which only had 10 billion parameters. Due to the vast amount of data analyzed, GTP3 is capable of mimicking human conversation to a high degree. YouTube videos of AI talking sounds like just another human being in front of the camera.
So how it apply in our daily lives?
Even prior to GPT3, there was AI that could, based on a few snippets of data provided by the user, generate entire chapters and even write novels. GPT3 can do all that, except, now it’s better, faster and more efficient.
In theory, any job that requires language in order to perform tasks can be accomplished, or at the very least assisted, through GPT3. Blogs and article entries, creative writing, and chat-bots are just some of the ways that GPT3 may impact our day to day lives, through some basic verbal input.
But the true beauty of GPT3 is when its paired along with other systems in order to enhance them. For instance, assuming that a process has both a linguistic and physical component, GPT3 may be used to perform communications flawlessly, an alternate AI may drive the physical, and another AI could be used to oversee the entire process ensuring that everything works according to plan. While this is a gross understatement of how complex in reality such a process would be, it goes to show that just like with most new innovations in the field of AI work best when paired with other systems.
Where do we go from here?
Currently the GPT3 is yet to be fully made available to the public, with many businesses that have received access to it already wait-listing customers for their services. An example of this would be Debuild, a tool that gives both professional web developers and non-web developers the opportunity to design and build functional web apps through simple linguistic input.
There are arguments against the use of such technology. Job displacement is one of the biggest arguments against artificial intelligence in general. However, counter arguments can be made by viewing it in the same light as the industrial revolution of the early 1800’s. While that transition was by no means easy, more jobs were eventually created as a result of improved productivity that offset job losses as a result. However, only time will tell whether or not GPT3 other future technologies will prove to be a detriment or an improvement to the world.
Below is a link to an actual interview with GPT3 performed by Eric Elliot in which you can truly understand just how powerful the AI is. Throughout the interview you can see that not only is the AI able to fully grasp the more open ended questions, but is also capable of sarcasm, predicting follow up questions from Eric based on where the conversation is leading and answering more philosophical questions such as “What is the meaning of Life”. We highly recommend you give it a watch as it shows just how far we as humanity has come through its tech aspirations.
Interview Between GPT3 and Eric Elliot
Read more In Technology
Artificial intelligence as a term was first coined by John McCarthy in 1956 where he held the first academic conference about the topic. While the concept of AI was previously discussed in theory years before McCarthy’s presentation, 1956 was the year that truly shifted the conversation from generalizing about the concept to fully evaluating its potential, threats and benefits to society.
In 2021, artificial intelligence is ever present, without you even noticing it. Amazon’s recommendations based on prior purchases, algorithmic trading and chat bots are some of the many implementations of AI in today’s society.
In this article we discuss GPT3, a neural network machine that has been trained via millions of data sets on the internet to read, comprehend, analyze and generate text based on linguistic input. The network was designed by OpenAI, and is by far one of the most sophisticated text-generating machines available to the world to date.
What is OpenAI?
OpenAI is an artificial intelligence research laboratory founded by several individuals, most notably Elon Musk. The purpose of the company has always remained to create and research artificial intelligence for the benefit of all humanity.
OpenAI has multiple projects that it has developed over its lifecycle. As a for-profit company, many of their projects are available for businesses to use through their website or through direct contact. That means businesses can use the latest innovations from one of the leading corporations on AI around the world at any given time.
One of their latest projects, OpenAI Codex, is an AI system that converts natural language inputs into code, which means that, hypothetically, it is capable of enhancing the speed at which developers generate code. It also means that less technologically adept coders would be able to generate elements or components to add to their code through basic linguistic input.
What is linguistic input?
Humans communicate through a variety of languages, whether Arabic, English or French, for example, or any of the hundreds of languages and dialects from around the world. Language as we know it is a purely human concept that we have adopted for the purposes of communicating with each other. Computers however speak a different, unified language: binary, where two digits define the underlying message of every sentence through ones and zeroes.
(Photo: Brent Walsh)
All those ones and zeroes represent values that are linked to each character within the text. Through these ones and zeroes, a computer is able to take inputs made by a human and understand them in order to perform an action.
In short, while you may speak one language, the computer is only ever able to comprehend binary. Here’s where it starts to get interesting.
Inferring meaning from language: a very human skillset
While most people understand jokes, metaphors and infer meaning behind text, a computer has a significantly harder time understanding chains of text altogether. When talking about a “grey chair” you would think of perhaps a grey colored chair, possibly situated in some kind of a room, probably in a shade of grey that others wouldn’t have thought of. The blanks that each of us fills in, the room, the shade of grey and maybe the shape of the chair, are all things that we infer and is second nature to humans. To a machine, however, this is no trivial task.
This is natural language processing (NLP) fits into the picture. NLP is the process of taking natural language that humans use and processing it in ways that would make it understandable for a machine. Unlike “dumb AI” of the past, which would look for specific key words or sets of words, an AI using NLP may infer meaning behind text and make assumptions about meaning, which is a critical component to language. However, as with all artificial intelligence, large amounts of data must be amassed in order to train the algorithm.
This is why the internet has played such a key role in developing artificial intelligence. All of our Google searches, all inputs we make on websites that collect our data, everything we do has been gathered, sorted and fed to algorithms. This I show a machine learns how to read, understand and speak “human”.
Early iterations of artificial intelligence were primarily based on mathematical functions that were sought to interpret specific data into specific results. Modern AI is capable of taking a more holistic approach to language and its meaning, interpreting and inferring how language is used and underlying undertones in order to generate responses.
In other words, modern AI machines not only understand what we’re saying, but infer the hidden meaning behind our words, and uses this knowledge to produce highly relevant results.
How does this affect me?
Machines are able to go through a large amount of data in seconds, analyzing, determining and inferring meaning behind the language used within a text. Hypothetically, this means that anything we may request from an AI that is sufficiently trained and is capable of performing tasks, can be carried out on a whim.
Having machines understand the human language in its full context has been a challenge since the inception of the concept of AI back in 1956. For all intents and purposes, the perfect AI acts, behaves and speaks like a regular human being, with the added benefit of being hooked up to a large pool of data available for processing, in this case the internet.
In essence, a truly powerful AI that is capable of mimicking human behavior and understanding, is something that modern day scientists strive for more than anything else due its potential in changing the overall technological landscape. General Intelligence, or GI, is what modern day AI researchers consider machines that are capable of constructing and displaying cognitive human-like abilities.
When prompted to “Write this as an Attorney” and given only the first two sentences as prompts, the GPT3 managed to write this masterpiece. (Photo: Francis Jervis)
Humanity is still far away from the GI technology needed to write this article; however, every day we move closer to the inevitable point where humanity and machines will be able not only to communicate but understand each other at a level that we assumed only humans were capable of.
What is GPT3 and why does it matter?
Before we go into GPT3, we must first understand what a neural network is. A neural network is a computing network that has been developed in tandem with biological neural networks found within animal brains. In other words, its essentially the brain’s infrastructure as we know it implemented into the digital market place. It is used to educate, train and evaluate a computer’s output paired with inputs in order to further its knowledge on subjects.
GPT3 is one of the most complex natural language processing models in the world that has a natural language generation component. This component is capable of not just understanding concepts, both implicitly stated and inferred through text, but also provides highly relevant and contextual output in text form as well. It’s one step closer to a truly human machine.
Aside from working in tandem with GPT Codex to create functional code, GPT3 is capable of generating language outputs akin to that of a real human. Through the use of over 175 billion machine learning parameters, GPT3 is far more advanced than its predecessor, the Turing NGL Model, which only had 10 billion parameters. Due to the vast amount of data analyzed, GTP3 is capable of mimicking human conversation to a high degree. YouTube videos of AI talking sounds like just another human being in front of the camera.
So how it apply in our daily lives?
Even prior to GPT3, there was AI that could, based on a few snippets of data provided by the user, generate entire chapters and even write novels. GPT3 can do all that, except, now it’s better, faster and more efficient.
In theory, any job that requires language in order to perform tasks can be accomplished, or at the very least assisted, through GPT3. Blogs and article entries, creative writing, and chat-bots are just some of the ways that GPT3 may impact our day to day lives, through some basic verbal input.
But the true beauty of GPT3 is when its paired along with other systems in order to enhance them. For instance, assuming that a process has both a linguistic and physical component, GPT3 may be used to perform communications flawlessly, an alternate AI may drive the physical, and another AI could be used to oversee the entire process ensuring that everything works according to plan. While this is a gross understatement of how complex in reality such a process would be, it goes to show that just like with most new innovations in the field of AI work best when paired with other systems.
Where do we go from here?
Currently the GPT3 is yet to be fully made available to the public, with many businesses that have received access to it already wait-listing customers for their services. An example of this would be Debuild, a tool that gives both professional web developers and non-web developers the opportunity to design and build functional web apps through simple linguistic input.
There are arguments against the use of such technology. Job displacement is one of the biggest arguments against artificial intelligence in general. However, counter arguments can be made by viewing it in the same light as the industrial revolution of the early 1800’s. While that transition was by no means easy, more jobs were eventually created as a result of improved productivity that offset job losses as a result. However, only time will tell whether or not GPT3 other future technologies will prove to be a detriment or an improvement to the world.
Below is a link to an actual interview with GPT3 performed by Eric Elliot in which you can truly understand just how powerful the AI is. Throughout the interview you can see that not only is the AI able to fully grasp the more open ended questions, but is also capable of sarcasm, predicting follow up questions from Eric based on where the conversation is leading and answering more philosophical questions such as “What is the meaning of Life”. We highly recommend you give it a watch as it shows just how far we as humanity has come through its tech aspirations.
Interview Between GPT3 and Eric Elliot
Read more In Technology
