Chat Generative Pre-trained Transformer (ChatGPT) was introduced on November 30, 2022, by OpenAI, a "research and deployment" company. OpenAI's website states that their mission is to "ensure that artificial general intelligence benefits all of humanity." Furthermore,  the company strives to build "safe and beneficial AGI," and they will deem their mission fulfilled if their "work aids others to achieve this outcome." Currently, ChatGPT exists as both a website and a Chrome extension.

ChatGPT was not the company's first foray into artificial intelligence; in January 2021, it introduced DALL-E, later upgraded to DALL-E 2. DALL-E 2 generates artistic or realistic images based on a text description provided by the user, sometimes modifying an original work or copying the style of a particular artist. Human artists were outraged, having spent years or even decades working on their craft. When Ammaar Reshi, a California-based product design manager, used ChatGPT to write a children's book in a weekend, children's authors were similarly enraged. (Reshi used another AI, Midjourney, for his illustrations.) The argument against Reshi and others who have created picture books using AI is that AI learns its skills from real-life examples, thus stealing from the writers and illustrators from which it learned. 

The education community also reacted strongly. Many believed "this is paradise for the cheating student." At the end of 2022, Furman University philosophy professor Darren Hick was quoted as warning of the "flood" of ChatGPT essays to come. Programs designed to detect content written by ChatGPT followed rapidly, including this one by Prince University student Edward Tian. 

ChatGPT is programmed to learn from conversations as it interacts with the user. It is able to "answer followup questions, admit its mistakes, challenge incorrect premises, and reject inappropriate requests." Its AI older sibling, InstructGPT, was designed to provide a detailed response to a prompt, just as a student would write a response to a prompt provided by an instructor, and like a student it learned from user feedback. A January 2022 report indicated that InstructGPT had significant limitations, including not having eliminated its ability to  "generate toxic or biased outputs, make up facts, and generate sexual and violent content without explicit prompting." It also displayed "bias "towards the cultural values of English-speaking people," a fault its creators were aware of and intended to improve upon before ChatGPT was made available to the public. As of the date of its release in November 2022, ChatGPT still lacked the ability to ask clarifying questions and, while efforts had been made to force the model to "refuse inappropriate requests, it will sometimes respond to harmful instructions or exhibit biased behavior."

The current version of ChatGPT still has weaknesses. For example, the information it knows dates from 2021 and before and has not been updated. Its creators acknowledge this limitation and others. The November 2022 version made available for free was ChatGPT3.5, and ChatGPT4 is anticipated this spring. CEO Sam Altaman tweeted on December 10, 2022, "ChatGPT is incredibly limited, but good enough at some things to create a misleading impression of greatness...[and it's] a mistake to be relying on it for anything important right now.” He then responded to a query, admitting his product to be as fallible as humans: "it does know a lot, but the danger is that it is confident and wrong a significant fraction of the time." In other words, user beware.

"artificial-intelligence-503593_1920" by alansimpsonMe is marked with Public Domain Mark 1.0.