OpenAI unveils GPT-4
Can handle image inputs and more lengthy requests
OpenAI has introduced GPT-4, which it says is more dependable, innovative and capable of processing significantly more complex commands than its predecessor, GPT-3.
Users have eagerly anticipated the release of GPT-4, which is built on a newer version of OpenAI's large language model (LLM) technology, since ChatGPT debuted in late November.
"We've created GPT-4, the latest milestone in OpenAI's effort in scaling up deep learning," reads OpenAI's blog post.
"GPT-4 is a large multimodal model (accepting image and text inputs, emitting text outputs) that, while less capable than humans in many real-world scenarios, exhibits human-level performance on various professional and academic benchmarks."
GPT-4's most notable feature is its capacity to process image inputs, in addition to textual ones. This means that the AI can now analyse images and generate textual responses based on its analysis.
GPT-4 has a significantly higher processing capacity than GPT-3, capable of handling up to 25,000 words: nearly eight times more than its predeccesor.
OpenAI trained GPT-4 on Microsoft Azure AI supercomputers, utilising Azure's "AI-optimised" infrastructure to enable worldwide access to the new model.
OpenAI claims that GPT-4 is not only more advanced, but also safer: it is said to be 82% less likely to generate responses that contain prohibited content, and 40% more likely to produce accurate and factual responses.
Additionally, OpenAI has released a list of tests that GPT-4 has passed.
The language model scored in the 88th percentile on the LSAT and in the 89th percentile on the SAT Math exam.
It also performed well on other standardised tests. It scored in the 80th percentile on the GRE Quantitative exam, and achieved good results on the GRE Verbal and Writing exams, scoring in the 99th and 54th percentile, respectively.
Despite the high expectations surrounding GPT-4, its founder, Sam Altman, acknowledged that the language model still has limitations and flaws, and that "it still seems more impressive on first use than it does after you spend more time with it."
Initially, GPT-4 will be exclusively accessible to ChatGPT Plus subscribers, who pay a premium fee of $20 per month. Additionally, the language model is being offered as an API for developers.
Microsoft's rapid integration of ChatGPT has ignited a competitive race among major tech companies, with Google announcing its own iterations of AI technology.
Other companies, including Amazon, Baidu, and Meta, have also entered the fray, striving to keep up with the latest advancements in the field.
There are growing concerns that generative AI technology, such as ChatGPT, may eventually replace many jobs currently performed by humans.
ChatGPT is capable of responding to queries using natural language, emulating human-like communication.
It has been used to generate various forms of creative content, such as songs, poems, and marketing copy, as well as to assist with tasks like coding and homework. Additionally, the AI can replicate various writing styles, including those of songwriters and authors, utilising the internet as its knowledge source based on data from 2021.
ChatGPT has already been integrated into Microsoft's Bing search engine platform, which is one of the many reasons why the tech giant has invested a staggering $10 billion into OpenAI.
Microsoft is slated to unveil additional AI products later this week at an event that is expected to focus on the future of work, with a particular focus on integrating new AI capabilities into Office and Microsoft 365.
These new features aim to enhance productivity and streamline workflows through the use of advanced AI technologies.