The hype surrounding ChatGPT continues, and after the latest announcement, AI is only going to get more people hooked. Because those interested in AI have been waiting for the new GPT version for years. Now everything goes in quick succession.
After the German Microsoft CTO Andreas Baum surprisingly announced the new language model GPT-4 for this week on March 9th, it was officially presented by OpenAI yesterday.
You can now find out what the new language model can do and what that means for you.
In this article you can find out what you can do with ChatGPT in everyday life and how helpful it is:
What is GPT 4?
GPT is the language model on which the popular text AI ChatGPT is based. It was previously based on GPT-3.5, but is now supposed to be significantly more powerful with GPT-4. The AI company says the model is “more creative and collaborative” and “solves difficult problems more easily” than ever before.
The differences to GPT-4 are rather inconspicuous in a normal conversation according to OpenAI. Especially with more complex tasks, the differences should be noticeable.
One of the biggest innovations: GPT-4 can react to text and image inputs. However, the language model should continue to only output text. Although this is not always supposed to be 100 percent at a “human level”, the language model is said to have been tested with many professional and academic benchmarks.
The US bar exam, for example, passed the AI and was in the top ten percent of the participants. For comparison, GPT-3.5 was only in the bottom ten percent.
You can see how GPT-4 works compared to GPT-3.5 in the developer live stream:
Link to YouTube content
How do I use GPT-4 for myself?
The brand new language model will initially be usable via ChatGPT Plus and Microsoft Bing. ChatGPT Plus is OpenAI’s subscription model that allows you to use ChatGPT with quick answers and at any time for $20 a month.
Bing is Microsoft’s search engine that aims to compete with Google by being equipped with ChatGPT. So far, however, you have to be put on a waiting list if you can use the new Bing.
Also, like GPT-3.5, GPT-4 will be released as an API to give developers the ability to build on top of it. There is also a waiting list for this. Since yesterday (03/14/23) interested parties have been invited one by one.
link to Twitter content
Does GPT-4 have weaknesses?
Yes, even the new language model is not without flaws and weaknesses. OpenAI points out that the language model still tends to invent information or write violent texts. Anyone who has ever used ChatGPT knows that this doesn’t happen too often if you don’t provoke it.
The ability to identify misinformation has arguably increased quite a bit compared to GPT-3.5. OpenAI speaks of an improvement of 40 percent.
The German Microsoft CTO Andreas Baum had indicated in his spontaneous announcement a few days ago that GPT-4 videos could be output. However, OpenAi has said nothing about videos in connection with ChatGPT or GPT-4.
This does not mean, however, that something in this direction could not come.
What do you think of GPT-4? Will you try using Bing or ChatGPT Plus? Write it to us in the comments!