In July 2020 the MIT Technology Review announced the development of a new language generator A.I. called GPT-3 and produced by OpenAI, a research lab founded by Elon Musk and Peter Thiel. Because of its vast neural network, GPT-3 performs much better than its predecessors being able to write short stories, songs, press releases, technical manuals, computer code, but also imitate established writers, and translate to and from a variety of languages. GPT-3 even wrote an informative article about itself that one would not be able to guess it wasn’t written by a human.
Here’s an example of GPT-3 creative writing: “When you can look into the mirror and see a poem looking back at you. When you can hear music in the play of silence. When you can create a writing that leaves people stunned. When you can laugh and weep as you think and breathe and bleed and eat and sleep. When you can dream with the quill in your fingers, then perhaps you will be a poet, a Poet, an Uber Poet.”
GPT-3 is good at synthesizing vast amounts of data and generating text on demand, but is also erroneous at times, and even racist and sexist. Having been trained on a dataset of half a trillion words, GPT-3 is able to identify an array of linguistic patterns but doesn’t understand what the words really mean and also lacks a general sense of purpose and meaning.
Given its versatility but also shortcomings, it remains to be seen how GPT-3 is going to be used by both academia and the business world to develop applications.