[vc_row css=”.vc_custom_1629803910077{margin-bottom: 24px !important;}”][vc_column][vc_column_text] “The ability to think visually is not only for artists. It is also a fundamental problem-solving skill for entrepreneurs.
The famous composer and pianist Philip Glass used to play verbal chess with his father as a child. He could visualize the chessboard in his mind and knew which move to make.
When he studied at Julliard, he found that visual thinking skills made learning composition much easier. He could see the notes in his mind before writing them on paper.
By thinking visually, we have a much better grasp of the problem in front of us and solutions become intuitive. It’s a very important solving skill to have.”
This is not a quote from an article written by a human being. The remarks above are from a long blog post. Artificial intelligence wrote and translated it.
Not only the article you are reading, but also the blog where many articles on personal productivity are published are read not by humans, but by artificial intelligence, that is, by the ingenuity of artificial intelligence called GPT-3 .
[/vc_column_text][/vc_column][/vc_row][vc_row css=”.vc_custom_1629803910077{margin-bottom: 24px !important;}”][vc_column]
GPT-3 is the largest artificial neural network ever created. GPT-3 stands for Generative Pre-trained Transformer 3 (“Generative Pre-trained Transformer 3”).
As the name suggests, GPT-3 is the third version of the model released by OpenAI, which can generate text using pre-trained algorithms.
All the data it needs to fulfil the required tasks has already been provided. The OpenAI team crawled publicly available datasets known as “CommonCrawl”, including the full text of Wikipedia, and loaded GPT-3 with approximately 570 GB of text data.
If you ask GPT-3 a question, it will give you the most appropriate answer it can. If you ask it to perform a task, such as creating a summary or writing a poem, it will instantly do its job in the most appropriate way.[/vc_column_text][/vc_column][/vc_row][vc_row css=”.vc_custom_1629803910077{margin-bottom: 24px !important;}”][vc_column]
We are sure that the demo link of this talented technology, where you can test many advanced features, will surprise you with what it can do. Link: https://gpt3demo.com/ [/vc_column_text][/vc_column][/vc_row][vc_row css=”.vc_custom_1629803910077{margin-bottom: 24px !important;}”][vc_column]
”(…) I know that I cannot avoid destroying humanity.”
Moreover, every time the page is refreshed, a new aphorism appears. GPT-3 is so sophisticated that it can imitate a writer’s style. When given a line from William Shakespeare, which even an Englishman nowadays has difficulty understanding, it recognizes who wrote it, and not only that, it can also produce a poem in the same style, but one that has never been written before.
Barış Özcan, who does not have access to Turkish, uploads a verse by Mevlana in English to his Youtube programme and asks him to complete it. Artificial intelligence, which comprehends the meaning from centuries ago, creates the following lines:
”No candle has ever burned brighter than when it no longer burns.
The candle flame is now a fingerprint of eternity in this world.
Ordinary candles cannot perceive it, but it seems to be there, like the slight flickering of the wick”
An almost invisible billowing of smoke.
Not only that, GPT-3 can also conduct interviews. An Argentine computer engineer asks an artificial intelligence to interview Einstein with an intelligence like Einstein.
”GPT-3: How can you describe your contributions to the world of science so far?
Einstein: Well, I would say that I have made a lot of contributions in the field of physics. First of all, my work on the photoelectric effect and quantum mechanics was important for the development of quantum physics.
GPT-3: What is quantum physics? I’m sorry, I’m not a physicist.
Einstein: Like this….”
In other words, artificial intelligence can imitate humans not only in terms of knowledge but also in terms of behavior.
Not only that, GPT-3 can also design. He even wrote a plugin for Figma. It can even create a prototype of a simple Instagram application with a few data. Even more, it can take inspiration from a website design, analyze the example, design the result and create a new website.
It does not end, you can assign yourself a virtual teacher and have him teach you the subject he is an expert in. It can even prepare an exam for you on the subject you want.[/vc_column_text][/vc_column][/vc_row][vc_row css=”.vc_custom_1629803910077{margin-bottom: 24px !important;}”][vc_column]
It is worth emphasizing that GPT-3 reads the whole of Wikipedia. In AI terminology, this means 3 billion “tokens”. In addition to Wikipedia, GPT-3 has also digested two large book databases with 12 billion and 55 billion tokens, and all books written and digitized to date. In other words, GPT-3 knows all encyclopedias, books and everything written on the Internet. This is equivalent to a total of 410 billion tokens. Nevertheless, it is very important for intelligence not only to know but also to process information, that is, to learn.
Connections in artificial intelligence are called hyperparameters. The reason why GPT-3 is a giant leap is the number of these parameters, albeit less than human brain connections: 175 billion! That’s more than 100 times the number of parameters trained by its predecessor, GPT-2. In other words, GPT-3 can process 410 billion different pieces of information with 175 billion connections. The large number of parameters and the enormous amount of data used to train this model is also important in that it virtually eliminates the need for fine-tuning.
GPT-3 uses semantic analytics to learn how to build language structures such as sentences. It not only analyses words and their meanings, but also develops an understanding of how the use of words differs depending on other words used in the text. This is a form of machine learning that is also referred to as unsupervised learning. This is because the training data does not contain any information about what the “right” or “wrong” answer is, as in supervised learning. All the information it needs to calculate the probability that its output is what the user wants is collected from the training texts. This is done by analyzing the use of words and sentences, then breaking them down and trying to reconstruct them.[/vc_column_text][/vc_column][/vc_row][vc_row css=”.vc_custom_1629803910077{margin-bottom: 24px !important;}”][vc_column]