Microsoft has announced that ChatGPT-4, the newest and most advanced release of the artificial intelligence model, will debut next week. It’s a bit strange that this announcement came casually during a Microsoft event and not from the OpenAI company itself, but Microsoft executives should certainly be familiar with ChatGPT’s development plans: it is the largest and most important partner in the project, and in fact finances it.
During the same event, it was said that ChatGPT-4, beyond improved capabilities in decoding and writing text, will also support other types of media. It seems that the technology will also be able to produce videos and possibly even audio clips based on text, and will improve OpenAI’s capabilities for turning text into speech and vice versa.
However, at the same time as these announcements, it seems that OpenAI is actually dealing with the recession of expectations from ChatGPT-4. During an interview with the company’s CEO, Sam Altman, he specifically stated that “people plan to be disappointed by the technology and it will indeed disappoint them”, when he means that many believe that ChatGPT-4 will bring with it a historic leap in artificial intelligence capabilities.
As for some of those prophecies, such as the ability to decode 100 trillion parameters instead of 175 billion in ChatGPT-3, Sam states that this is simply “complete bullshit”, as is the case with most of the other predictions. “The rumors surrounding GPT-4 are just ridiculous, I have no idea where this is coming from,” he said.
Currently, ChatGPT-3 is the public and open version of the technology, while ChatGPT-3.5 serves the Bing search engine.