One-stop-shop to get information into the history, development and potential of GPT-3.
Julien Lauret’s article is a comprehensive summary of the journey taken so far to create GPT-3.
Julien has managed to summarize years of development and introductions of methodology and techniques to model language and solve natural language processing into several small, concise paragraphs.
As well as providing the reader with a background of GPT-3, Julien also gives a somewhat diplomatic answer to the question as to whether GPT-3 is AGI. His response truly reflects the nature of the question itself, in that the question is subjected to the definition of intelligence by whoever poses the question.
Julien’s latest article on GPT-3 is a well-needed article, as I feel the topic of Artificial General Intelligence(AGI) has almost taken a backseat in our imagination.
Julien points early in the article, two viewpoints on AGI. One viewpoint alludes to the fact that AGI is decades away; while the other questions the possibility of humans ever achieving AGI.
Julien’s article is littered with some technical and specialist terms, but each term is accompanied with a brief definition and in some cases an extensive background.
Terms such as ‘Deep Neural Networks’, ‘Machne translations’, ‘Word2vec’ and ‘Few-shot learning’ are presented and defined in a manner that provides readers with a crash course on language modelling and NLP, as well as the primary information around GPT-3.
The article doesn’t mention GPT-3 until halfway through the article, due to the fact that Julien cleverly takes the reader through a journey that outlines crucial developments and research that have contributed to the advancement of language modelling and GPT-3 itself.
Information on the intrinsic characteristics(number of weights) of both GPT-3 and its predecessor are provided to give the reader an understanding of the level of progress the GPT model has made at every released version.
The latter half of the article presents examples of the intuitive application of GPT-3, developed by early access API users.
But more importantly, Julien answers the question proposed in the article’s title. Julien’s conclusion at to whether if GPT-3 is AGI falls on the conclusion that it isn’. But even more so, the answer to the question itself is not as clear as one might think.
The unclarity stems from the fact that the definition of intelligence is ambiguous and subject to an individual’s interpretation.
What Julien states that is an important matter to discuss is the application of GPT-3. We’ve already seen its ability to generate poems, gameplay scenarios and web components defined through language.
What we now need to observe is how the latest feat of AI can be applied in a more general setting that involves videos and images.
Julien’s article interestingly contains segments of philosophical views and statements. For example, Julien points to the proposition that we can’t determine the level of intelligence GPT-3 possesses, just as we can’t determine the level of cognitive function of a paralyzed person or individual with physical limitations.
This article delves touches areas of AI that are worth pondering about.
This article is a great read for:
Machine Learning Practitioners: The language and technical term used is very familiar with anyone involved in machine learning. And the definitions and papers included allows for further exploitation of topics related to GPT-3