Inside OpenAI’s empire: A conversation with Karen Hao

The third advantage is that the empires monopolize knowledge production. Therefore, in the past ten years, we have seen the artificial intelligence industry monopolizing more and more researchers of artificial intelligence in the world. Therefore, artificial intelligence researchers no longer contribute to opening science or work in universities or independent institutions, and the influence on research is what he may imagine will happen if most climate scientists in the world are transferred by oil and gas companies. You will not get a clear picture, nor do we get a clear picture, the restrictions of these technologies, or if there are better ways to develop these technologies.

The fourth and final feature is that the empires always participate in this aggressive race, as there are good empires and evil empires. And they must be, the righteous empire, to be strong enough to overcome the evil empire, and for this reason they should have an unrestricted license to consume all these resources and exploit all this work. And if the evil Empire gets technology first, humanity goes to hell. But if the righteous empire gets technology first, they will embrace the world, and humanity will go to heaven. So on many different levels, such as the subject of the empire, I felt that it was the most comprehensive way to name exactly how these companies work, and what exact effects on the world.

Niall Verth: Yes, wonderful. I mean, you are talking about the evil empire. What happens if the evil Empire gets it first? What I mentioned above is AGI. For me, it is almost like the additional personality in the book all the way. It is a kind of waving on the horizon on everything, like a ghost on Eid, a kind of saying like, this is the thing that stimulates everything in Openai. This is the thing that we should reach before anyone else reaches it.

There is some in the book about how to speak internally in Openai, for example, we have to make sure that AGI is in our hands where it is safe for anywhere else. Some international employees are similar to publicly – this is a kind of strange way to frame it, right? Why is the American version of AGI better than others?

So he told us a little about how to set what they are doing. And AGI is not an inevitable fact that occurs in any case, right? It is not something yet.

Karen Hao: There is no consensus on whether or not it is possible. There was recently New York Times A story by Cade Metz, which was referring to a survey of artificial intelligence researchers for a long time in this field, and 75 % of them still believe that we have no techniques yet to reach AGI, whatever that. The definition or the most classic understanding of what is AGI is the ability to completely create human intelligence in software. But the problem is that we also do not have a scientific consensus on what human intelligence is. Thus, one of the aspects that I am talking about in the book is that when there is a vacuum of the common meaning about this term, and how will it look like, when we have reached it? What are the capabilities we should evaluate these systems to determine that we have come there? It can only be what Openai wants.

Therefore, it is a kind of permanent goal that continues to shift, depending on where the company wants to go. As you know, they have a full set, and a variety of different definitions they used over the years. In fact, they have an internal joke: If you ask 13 researchers from Openai what is AGI, you will get 15 definitions. So, she realizes themselves that this is not a true real term and has no meaning.

But it serves this purpose of creating a kind of semi -religious enthusiasm about what they do, as people believe that they have to continue to drive towards this horizon, and that one day when they arrive there, it will have a transformational effect of civilization. Thus, what should you do in your life, but this? Who should work on it, but you?