OpenAI, the company behind Dall-E and ChatGPT, has announced the successor to its smart algorithm: GPT-4. OpenAI has now attracted considerable attention and Microsoft, among others, is busy with a GPT-AI language to improve its Bing search engine. However, a new language variant has now been announced, namely GPT-4.
OpenAI has focused on creativity. It also mentions that it needs to work better together and that more difficult problems can be solved with this new GPT. It would also be more accurate, which ChatGPT can sometimes use, for example. OpenAI is honest about that, by the way. It says the system could certainly suffer from the same problems as previous language models and that, among other things, it “hallucinated” information together.
But this new one GPT-4 brings something new to the table. For example, it is not only able to handle text input: it can also handle photo input. However, it only answers in text. GPT-4 is already available, namely for people who have ChatGPT Plus. This subscription costs $ 24.20 per month if you purchase it from the Netherlands. GPT-4 is also already built into Microsoft’s search engine.
According to OpenAI, the differences between the previous model (GPT-3.5) and this one are not huge. However, it is especially impressive when you use it for the first time. But hey, that goes for almost all artificial intelligence tools. However, the company has tested the artificial intelligence equally thoroughly by having it take the CITO test (at least, the American version of it for slightly older children, the SATs). He scored a percentage of 88 percent and higher, which means that it is still a very smart system. We are therefore not surprised that companies such as Duo Lingo and Stripe are working to integrate GPT-4 into their apps. Especially when you see the examples of how well ChatGPT understands what can be seen in an image. It can be called bizarre.
Due to the recent popularity of ChatGPT, it seems quite normal that OpenAI updates GPT, but that is not the case. There has been speculation for a year about what GPT-4 could do. OpenAI already said that people have far too high expectations of it. Although the developments of artificial intelligence seem to be moving very quickly, and indeed the focus is very much on it now, that does not mean that innovations will now follow each other at a rapid pace. Now we have seen that Google seems to be speeding up its AI, but at the same time, not much has come of the recently announced new options such as in Workspace and Bard. This is all still in a testing phase.
That is not a criticism: artificial intelligence can have a major impact on human lives and should therefore be handled with care. But that it is the talk of the day in the tech world, that is one thing that is certain. This also means that expectations of new announcements from companies such as Google and OpenAI are high, while those companies mainly want to focus on making the AI more reliable. It is not sexy, but it is essential if we want to continue with AI in the form that we have been able to experience so far and in forms that we have yet to experience.