ChatGPT: What is ChatGPT, how does it work, what response did it get from the public and who made it?

The service was developed as free to the public in the beginning, with plans to monetise the programme later down the line. By December 4, 2022, it already had an estimated one million plus users and this number grew to more than 100 million in January 2023, making it the fastest growing consumer application to date.

What is ChatGPT and who developed it?

ChatGPT is a language model trained to give human-like responses to questions, admit its mistakes, challenge incorrect information and reject inappropriate questions.

It was developed by OpenAI, an American artificial intelligence research laboratory, owned by CEO Simon Altman. It is built on top of OpenAI’s GPT-3 family of large language models and has been modified using both supervised and reinforcement learning techniques.

When it first launched, ChatGPT had quickly attracted attention for its detailed responses and articulate answers across many domains of expertise. However, its uneven factual accuracy was seen as a significant drawback.

A valuation conducted by OpenAI was estimated at £24,000,400,000 ($29,000,000,000).

How does ChatGPT work?

Whilst the main purpose of a chatbot is to impersonate a human conversationalist, ChatGPT is a lot more adaptable. For instance, it can write and debug computer programmes, compose music, teleplays, fairy tales, and student essays, it can answer test questions, write poetry and song lyrics, emulate a Linux system, simulate an entire chat room, play games and simulate an ATM.

Its training data includes ‘man pages’ and information about internet phenomena and programming languages, like bulletin board systems and the Python programming language.

Unlike its predecessor, InstructGPT, ChatGPT attempts to reduce harmful and deceitful responses.

Compared to most chatbots, this version remembers previous prompts and questions given to it in the same conversation. Journalists have suggested that this will allow ChatGPT to be used as a ‘personalised therapist’.

To avoid offensive prompts from being presented to and produced from ChatGPT, questions are filtered through OpenAI’s moderation API, and potentially racist or sexist prompts are dismissed.

What responses did ChatGPT get?

The service received a mixed response of positive and negative reviews.

In December 2022, it was met with some positive reviews; New York Times journalist Kevin Roose named it “the best artificial intelligence chatbot ever released to the general public”, The Guardian reporter Samantha Lock said that it was capable of generating “impressively detailed” and “human-like” words.

Technology writer Dan Gilmor used the service on a student assignment and discovered that it curated text nearly as good as the work a good student would deliver and said that “academia has some very serious issues to confront”. In relation to the service’s ability to debunk inaccurate information, Slate magazine journalist Alex Kantrowitz praised ChatGPT’s pushback to questions related to Nazi Germany, such as the statement claiming that Adolf Hitler built highways in Germany.

However, the programme was also faced with widespread criticism from educators, journalists, artists, ethicists, academics and public advocates. The Verge journalist James Vincent believed that the rapid success of ChatGPT showed that artificial intelligence had gone mainstream.

Journalists also commented on ChatGPT’s proneness to ‘hallucinate’; Mashable blogger Mike Pearl tested the service with multiple questions, when he asked for ‘the largest country in Central America that isn’t Mexico’, ChatGPT responded with Guatemala, when the answer is in fact Nicaragua. Another inaccuracy in the system was when CNBC asked ChatGPT for the lyrics to ‘The Ballad of Dwight Fry’, the service responded with made-up lyrics rather than the actual lyrics.

In December 2022, Stack Overflow, a question and answer website, banned ChatGPT due to its factually inaccurate answers to prompts.