OpenAI releases ChatGPT

An artificial intelligence company in San Francisco has released a chatbot that can answer questions, write essays and compose poems, and it has quickly attracted huge public interest.

ChatGPT was released by OpenAI on 30 November as a free research preview. The company's chief executive, Sam Altman, said on 5 December that it had passed one million users.

How it works

ChatGPT is built on a large language model, a type of artificial intelligence trained on vast amounts of text from the internet and books. It generates responses by predicting which words are likely to come next.

The chatbot has been refined using feedback from human trainers, which helps it to follow instructions and give more helpful answers. Users type questions or requests in ordinary language, and it replies in a conversational style.

What it can do

Users have been sharing examples of the chatbot writing computer code, explaining complex subjects, drafting emails, writing stories in the style of famous authors and answering exam questions.

Many have been impressed by how natural and fluent its answers are.

Limitations

OpenAI has warned that ChatGPT sometimes writes answers that sound plausible but are wrong or nonsensical. It can also be sensitive to the way questions are phrased, and its knowledge of events after 2021 is limited.

The company has built in safeguards to stop the chatbot producing harmful content, but users have found ways around some of them.

Concerns

Teachers and universities have raised concerns that students could use the chatbot to write essays and homework. Some experts worry that such tools could be used to spread misinformation or to produce spam on a vast scale.

Others have questioned what it could mean for jobs such as writing, customer service and computer programming.

The AI race

OpenAI was founded in 2015 as a research organisation, with backing from investors including Elon Musk, and has received billions of dollars from Microsoft. It is also known for DALL-E, a program that creates images from text descriptions.

Other technology companies, including Google, have been developing similar systems but have been cautious about releasing them to the public.

Many experts believe the release of ChatGPT could mark a turning point in how people use artificial intelligence.