Tuesday saw the release of the most recent iteration of OpenAI’s main big language model, GPT-4, which the company claims performs at “human-level performance” on several objective tests.
ChatGPT-4 is “larger” than previous versions, which means it has been trained on more data and has more weights in its model file, making it more expensive to run as well.
Many AI researchers today think that many recent advances in the field result from training ever-larger models on thousands of supercomputers, which can cost millions of dollars. GPT-4 is an example of an approach centering around “scaling up” to achieve better results.
OpenAI said it used Microsoft
Microsoft has invested billions in the startup and is using Azure to train the model. Citing “the competitive landscape,” OpenAI declined to disclose information about the precise model size or the hardware it used to train it.
The latest version of OpenAI’s GPT large language model is a preview of new developments that could start filtering down to consumer products like chatbots in the coming weeks. OpenAI’s GPT large language model powers many of the artificial intelligence demos that have been astounding people in the technology industry over the past six months, including Bing’s AI chat and ChatGPT. Bing’s AI chatbot uses GPT-4, Microsoft said on Tuesday.
OpenAI says the new model will produce fewer factually incorrect answers, go off the rails and chat about forbidden topics less often, and even perform better than humans on many standardised tests.
GPT-4 performed at the 90th percentile on a simulated bar exam, the 93rd percentile on a SAT reading exam, and the 89th percentile on the SAT Math exam, OpenAI claimed.
However, OpenAI warns that the new software isn’t perfect yet and that it is less capable than humans in many scenarios. It still has a major problem with “hallucination,” or making stuff up, and isn’t factually reliable, the company said. It is still prone to insisting it is correct when it is wrong.
The business stated in a blog post that it is working to address the “many acknowledged limitations” of GPT-4, including social biases, hallucinations, and antagonistic prompts.
“In a casual conversation, the distinction between GPT-3.5 and GPT-4 can be slight. When the task’s complexity reaches a certain level, GPT-4 distinguishes itself from GPT-3.5 by being more dependable, inventive, and capable of handling much more nuanced instructions, according to a blog post by OpenAI.
The new model will be accessible to ChatGPT users who have paid for it as well as through an API that enables developers to incorporate the AI into their applications. For every 750 words of prompts, OpenAI will charge roughly 3 cents, and for every 750 words of response, they will charge about 6 cents.