One of the most popular AI powerhouses worldwide, OpenAI launched and unveiled the next iteration of its Generative Pre-trained Translation model (GPT), GPT-4. The technology has become far bigger, much faster, and more accurate than the current version, ChatGPT. The company says the language model has become more collaborative and creative. The significant advancements and updates from ChatGPT’s foundational GPT 3.5 in this version include the ability to process images and generate text or captions post-analysis of the prompt image.
The parent company, OpenAI, says that this new iteration of GPT exhibits human-level performance on various academic and professional industrial benchmarks. The new language model can perform well and pass a bar exam with a similar score as scores by the top 10 percent of humans participating in the exam.
The new model has updated capabilities to solve difficult problems with better accuracy. For instance, it can schedule a meeting among four busy people, answer tax-related questions or learn the creative elements from the writing style of any user. GPT-4 has become more rigorous than ever.
The updated text processing limit of over 25,000 words allows it to develop new use cases, including long-form content development and creation, document analysis, search capabilities, or extended conversations. Per a report from OpenAI, this new GPT has an 80% better potential to refrain from responding to disallowed content and a 40% better potential to produce factual responses than the current iteration of GPT.