GPT-4 Launches Today, But You May Have Already Tried It – Review Geek

Four months after the extremely successful release of ChatGPT, OpenAI is back with an updated language model. GPT-4 launches today and will soon be integrated into apps like Duolingo and Khan Academy. But there’s a decent chance you’ve already tried GPT-4.

In a new blog post, Microsoft says that its Bing chatbot already uses GPT-4. In fact, it was released with GPT-4. This is a bit surprising: Microsoft previously described the Bing chatbot as a “more powerful” and “customized” version of GPT-3.5. (To be perfectly clear, the Bing AI chatbot is not a basic version of GPT-4. The basic model of GPT-4 cannot access information beyond September 2021, among other differences.)

The GPT-4 debut is quite modest. This appears to be an incremental update, and the CEO of OpenAI Sam Altman warns that GPT-4 “looks more impressive on first use than after spending more time with it.” Based on our experience with the Bing chatbot, I am inclined to agree. Like GPT-3.5, the new GPT-4 model is prone to inaccuracies and hallucinations.

That being said, GPT-4 is still a remarkable advance. OpenAI claims that this new language model is “40% more likely” to produce factual answers than its predecessor. And interestingly, they are 82% less likely to provide “disallowed content” to the user. (But I’m a bit skeptical, as we didn’t have much trouble pulling the initial notice and other “disallowed content” from the Bing chatbot.)

And GPT-4 is a multimodal system, which means it can work with more than just text. But this multimodal system is a bit rudimentary. GPT-4 can “see” images, which should be incredibly useful, but it can’t. produces images (except for ASCII images). The model output is still limited to language.

I must reiterate that this is an incremental update. GPT-4 still lacks general intelligence and, in the words of CEO Sam Altman, “it’s still flawed, it’s still limited.” Exaggerating or exaggerating the capabilities of this AI is a bad idea: language models like GPT-4 are useful and interesting, but they shouldn’t be treated as an authoritative source, especially in environments like Bing or Google Search.

You can try GPT-4 today with ChatGPT Plus (with a limit of 100 messages every 4 hours). Or just join the Bing Chatbot Early Access Program. The multimodal functionality does not work at this time; presumably, it will release after the GPT-4 developer event on March 14 at 4pm ET.

Source: Open AI