Member-only story
OpenAI is anything but...
OpenAI has been trained on 45Tb of human output over the last few years but perhaps it hides a darker secret when it comes to the Orwellian nature of the system.
TLDR: The data set used whilst coming from a range of years is subject to the biases that have emerged over that time, driven by a media narrative the successor to which is OpenAI and its generative principle to propagate and heighten specific biases.
OpenAI is fun to chat with, and great at rewriting your emails so they sound more “business-like” but like anything internet related, you can find yourself going down a rabbit hole very quickly and worse, sometimes you might not know you’re going down that rabbit hole.
Question: Could OpenAI be used to propagate and embed ideals in society as a whole, to push ideas and narratives that are currently very popular, and to continue the drive towards a tasteless, soulless, frictionless eunuch of a Western world?
Let’s ask the algorithm of the hour itself:
ME: How big was your training data set?
OAI: I was trained on a diverse range of internet text, including websites, books, and social media platforms, with a total of approximately 45 terabytes of text data.
ME: How old is your oldest social media data set
