Microsoft-backed startup OpenAI began rolling out GPT-4, a powerful artificial intelligence model that supersedes the technology behind the hugely popular ChatGPT.
GPT-4 is “multimodal”, meaning it can generate content from both image and text prompts.
What is the difference between GPT-4 and GPT-3.5?
GPT-3.5 only accepts text prompts, while the latest version of the large language model can also use images as inputs to recognize and analyze objects in an image.
GPT-3.5 is limited to around 3,000 word answers, while GPT-4 can generate answers over 25,000 words.
GPT-4 is 82 percent less likely to respond to requests for prohibited content than its predecessor and performs 40 percent better on certain factual tests.
It will also give developers the ability to choose the style and depth of their AI. For example, GPT-4 can adopt a Socratic conversational style, responding to questions with questions. The previous iteration of the technology had a solid tone and style.
Soon, ChatGPT users will have the ability to change the tone and style of the chatbot’s responses, OpenAI said.
What are the capabilities of GPT-4?
The latest version has surpassed its predecessor on the US bar exam and the Graduate Record Examination (GRE). GPT-4 can also help individuals calculate their taxes, as shown in a demonstration by OpenAI President Greg Brockman.
The demo showed that it could take a photo of a hand-drawn mockup for a simple website and create a real website.
Be My Eyes, an app for the visually impaired, will provide a virtual volunteer tool powered by GPT-4 in its app.
What are the limitations of GPT-4?
According to OpenAI, GPT-4 has limitations similar to its previous versions and is “less capable than humans in many real-world scenarios”.
Inaccurate responses, known as “hallucinations,” have been a challenge for many AI programs, including GPT-4.
According to OpenAI, GPT-4 can compete with human propagandists in many areas, especially when working with a human editor.
It cited an example where GPT-4 made suggestions that seemed plausible when asked how to get two parties not to contradict each other.
OpenAI Chief Executive Officer Sam Altman said GPT-4 is “the most powerful and most aligned with human values and intent,” though “it still has flaws.”
GPT-4 generally lacks knowledge of events that occurred after September 2021, when the vast majority of its data was truncated. Nor does it learn from experience.
Who has access to GPT-4?
While GPT-4 can handle both text and image input, only the text input feature is available to ChatGPT Plus subscribers and software developers with a waiting list, while the image input feature is not yet publicly available.
The subscription plan, which offers faster response time and priority access to new features and improvements, launched in February and costs $20 per month.
GPT-4 supports Microsoft’s Bing AI chatbot and some subscription-level features of the language learning platform Duolingo.
© Thomson Reuters 2023
