The AI model is said to reason better than ChatGPT and can also use images as input. In their example they gave the model an image of a bunch of balloons and asked it what would happen if the strings were cut. GPT-4 knew that the balloons would fly away, without first being told that these were balloons, so it both figured that out, and knew the logical characteristics of such things in our reality.

The company says that the model is an ideal candidate for learning, and a case study on their site showed how Duolingo has incorporated it into their chat feature, basically giving learners a personal language tutor to converse with in freeflow conversation.

It’s currently available through ChatGPT Plus, and devs can sign up to a waiting list to gain access to the API. Does anyone else get the feeling that as these models keep improving it seems like a risky idea having it all under the control of a single company? Not sure what the solution is since the genie is already out the bottle.

https://openai.com/product/gpt-4
https://www.youtube.com/watch?v=–khbXchTeE