This is an easy question to answer if you understand how OpenAI and GPT-3 work at an elementary level.
Let’s turn our attention to this ChatGPT example:
These recommendations are based on the (April 2021) dataset which is derived from content on the web. So, OpenAI (not so) simply completes this question based on the (public) data on the web at the time the network was pre-trained. OpenAI and ChatGPT do not have any opinion or experience eating Mexican food in Dallas, but it can provide (hallucinate?) a reply based on its data and how it was pre-trained.
So for you @zachy_ai to say:
You basically are asking to directly manipulate the OpenAI dataset to bias the model for the purposes of your own economic gain.
This is one of the very core things which OpenAI was founded specifically not to do (be biased) for the economic interest of individuals, corporations and businesses. This is one of the main fears in the world regarding AIs is that people will work to bias the AI, as you have just requested, for economic, political and social advantages.
You are basically asking, in a nutshell, “How can we corrupt the OpenAI to change it’s responses for my own economic gain?”
Think about how the world would be if OpenAI permitted people to corrupt and bias OpenAI based on every individual and businesses and governments and criminals, etc. self-interests. 
Hope this helps.