Artificial intelligence research lab OpenAI on Tuesday unveiled the latest version of its impressive language software, GPT-4, an advanced image analysis and human speech mimicking tool that pushes the technical and ethical boundaries of artificial intelligence, which is growing at an incredible pace.
According to The Washington Post, an earlier model of OpenAI’s artificial intelligence, ChatGPT, has charmed and unsettled the public with its incredible ability to write human love letters, do homework, and even program with deceptive fidelity.
The GPT-4, on the other hand, is an advanced model that can generate text, recognize an image, and outline in words what is visible in the picture. In fact, if you show it a picture of a boxing glove hanging over a bouncy castle with a ball on one end and ask it what will happen if the glove falls off, the GPT-4 will say that the glove will fall on the bouncy castle and the ball on the other end of the castle will fly up.
The developers believe the technology could further revolutionize not only our work but also our everyday lives. But fearing abuse, they are delaying the launch of key features for the time being. Therefore, the version available now is text-only, but ChatGPT Plus subscribers can now use GPT-4 on a limited basis.
OpenAI pointed out that the new version still has many bugs, including making up nonsense, perpetuating social prejudices, and giving bad advice. It is also unaware of events that happened after September 2021, when its data was finalized, and “does not learn from its experience,” meaning users cannot teach it new things.
