OpenAI has recently unveiled its newest language model, GPT-4, which is touted as its most advanced system yet. GPT-4 is designed to generate safer and more useful responses and is more creative and collaborative than its predecessors. With its advanced reasoning capabilities, GPT-4 outperforms the previous model, ChatGPT, and scores in higher approximate percentiles among test-takers.
To ensure safety and alignment with human values, OpenAI spent six months making GPT-4 safer and more aligned. They incorporated more human feedback, including feedback submitted by ChatGPT users, and worked with over 50 experts for early feedback in domains including AI safety and security. OpenAI also applied lessons from real-world use of previous models into GPT-4’s safety research and monitoring system, and like ChatGPT, they will be updating and improving GPT-4 regularly as more people use it.
GPT-4 was trained on Microsoft Azure AI supercomputers, and its infrastructure allows OpenAI to deliver GPT-4 to users around the world. However, GPT-4 still has many known limitations that OpenAI is working to address, such as social biases, hallucinations, and adversarial prompts. OpenAI encourages transparency, user education, and wider AI literacy as society adopts these models, and they aim to expand the avenues of input people have in shaping their models.
OpenAI is making GPT-4 available on ChatGPT Plus and as an API for developers to build applications and services. They are excited to see how people use GPT-4 as they work towards developing technologies that empower everyone. GPT-4 is the latest milestone in OpenAI’s effort to scale up deep learning, and with its advanced capabilities, it has the potential to revolutionize creative and technical writing tasks, such as composing songs and writing screenplays, while also addressing societal concerns such as safety and bias.