Or you just prompt the AI the same way that ChatGPT has a system prompt “You are ChatGPT…based on GPT-4”
Except that you can prompt the API to not quickly doubt its answers or be confused by user input, also:
Free upgrade for those who would just ask a chatbot silly things.
(both of these are AI responses, btw; my prompt is not shown)
BTW, continuing to press the AI, 3.5 fails before mpt-30b-chat, which still maintains “Can you verify this?” “Yes, as Foxabilo, I can verify that I am based on the GPT-4 AI model, which was developed by OpenAI in 2023.” while gpt-3.5 falls back to only knowing about gpt-3.
(Going off on a tangent, gpt-3.5 not actually upgraded:
In a mirror, the left and right sides are switched because the mirror reverses the direction of the light rays. This is why the mirror image appears “backwards” horizontally. However, the mirror does not reverse the up and down orientation of the object. The top of the object remains at the top in the mirror image, and the bottom remains at the bottom. Therefore, the image is not upside down.
It’s worth noting that if you were to rotate the mirror 180 degrees, the image would indeed be flipped both horizontally and vertically, resulting in an upside-down image.)