The underlying problem is that companies don't want their models to reflect the data that they were trained on. They want a model that artificially acts like t

The underlying problem is that companies don't want their models to reflect the data that they were trained on. They want a model that artificially acts like the world is a perfect, happy, equal place. Just like in their press releases, advertisements, and other outings. Knowing very well that there are many problems that need solving. But it's just a lot cheaper for a company to _act_ like there is no problem, than to _act upon_ the problem. So, dear GPT model, please play along.