OpenAI announced an updated version of its artificial intelligence technology that powers ChatGPT on Tuesday, called GPT-4.
The new text-based AI model, which is available for paid users of ChatGPT Plus, is more advanced than the tech that currently power ChatGPT. It ”surpasses ChatGPT in its advanced reasoning capabilities,” and scores higher than ChatGPT on tests like the bar exam and Biology Olympiad, OpenAI said in an announcement.
GPT-4 can also:
- Write code in all major programming languages
- Understand and analyze images as input from a user
- Read and analyze up to 25,000 words of text
OpenAI said GPT-4 is safer and more factual than previous GPT iterations: It’s 82% less likely to respond to requests for content that’s not allowed on the service, and 40% more likely to produce factual responses to prompts.
Still, the company said it has “many known limitations that we are working to address,” including social biases, adversarial prompts, and hallucinations, which is when a chatbot responds confidently with an incorrect answer that’s not based on its training data.
OpenAI has been working on GPT-4 since before the release of ChatGPT, which became a viral sensation after it was released late last year.
It sparked a race among tech giants to embrace and promote their AI-based services: Google announced a new chatbot based on its machine learning tech, and Chinese tech giant Baidu said it was working on one, too.
Microsoft, which invested billions into OpenAI, launched a limited release of its search engine Bing integrated with OpenAI's technology last month.
Microsoft is set to make another AI-related announcement on Thursday. In advance of that announcement, rival Google said Tuesday that it plans to include AI in its Google Workspace products including Gmail, Google Docs, and Sheets.