Open AI presents the next version
Open AI is currently holding its chatbot ChatGPT, who can answer all kinds of questions with the help of artificial intelligence (AI), has the technology world in suspense. The system is based on the GPT-3.5 language model. Rumors have been circulating in the industry for months about a successor version, GPT-4, which is said to be much more powerful.
Open AI officially presented this system on Tuesday, and the company described it as its “latest milestone”. It turned out that GPT-4 has been in use for weeks. Because the software company Microsoft, a close partner of Open AI, has now confirmed that the new version of its search engine Bing, which was launched in February, runs on GPT-4. So far, the company had only said that it was using an advanced version of ChatGPT.
The most striking difference between GPT-4 and GPT-3.5 is that the new system can not only handle text, but also images. Both text and images can be entered into the program, with the response then being text only. As an example, a photo of the inside of a refrigerator was mentioned in a blog entry. The system can not only identify the content, but also analyze what can be cooked with it. However, this image function is not yet widely available, there is currently only a waiting list for it.
Premium subscription for $20
However, the usual text functionality of GPT-4 is now made available through ChatGPT for users of the premium paid subscription, which costs $20 a month. Initially, the usage volume of the new system is to be limited.
Open AI describes GPT-4 as a significant improvement over GPT-3.5. While the new program is “less capable than humans in many real-world scenarios,” it demonstrates “human-level capability” in various professional and academic fields. For example, it passed a legal state exam and was in the top 10 percent. GPT-3.5, on the other hand, was in the bottom 10 percent. In simple conversations, the difference between GPT-4 and GPT-3..5 can be “subtle”. However, there is a noticeable deviation from a certain complexity of the task. Then GPT-4 is more reliable and also more creative.
As Open AI admits, GPT-4 still has “similar limitations” to previous language models. It’s prone to “hallucination,” meaning it gives wrong answers, albeit fewer than previous versions. And like GPT-3.5, the new model is not up to date and mostly only uses data up to 2021. It also does not learn “from its experience”.