I wonder what the interview process must be like. Who made the first move? OpenAI? Or the researchers? And how many rounds are there?
The more important the position, the more relaxed and casual the interview process, in my experience. I wonder at what point in a SWE's career (I am not a SWE) interviews go from "Solve this Leetcode problem to prove you are worthy" to "We would be honored if you would join us."
It erodes Google's position as the premier research institute.
If an ambitious grad wanted to make their mark, they'd either apply to one of Google's institutions (DeepMind, Brain) or be generated by TPUs (if they were literally a grad[ient]).
Nowadays, not so much. Researchers are realizing that Google is the place to go if you don't want your research to be incorporated into products, except indirectly (Youtube recommendations, google ad placement, etc). Meanwhile OpenAI is reshaping the world.
It's in OpenAI's interest to strike while the iron is hot, as they say. This hype will die down eventually, but displacing 30 million-dollar researchers from Google will have long-term effects that are hard to measure.
Plus, PhDs are the resource that can't be replaced. If someone magically vanished OpenAI's PhDs tomorrow, it would be very hard for them to recover. Whereas if someone vanished every remaining Twitter engineer, Musk could probably pull off a recovery by hiring replacements.
That's a bold assumption. I'd assume that very few people would sign up for a job under a dude who basically just lit $30B on fire and told his employees to sleep in the office - especially if they had other options.
It's easier to replace engineers than it is to replace researches, but it's definitely not easy.
It seems like, at least initially, FB and Google had far superior tech and talent compared to OpenAI. But they were slow to move, and with the recent media hype and the giant support from Microsoft, the best people seem to be jumping ship quickly over there now. I wonder how this will play out.
People said the same thing about cloud computing and machine learning. When the cost per compute approached commodity status for both, those opinions quickly changed.
People will say whatever the status quo says or they’ll be booted off the island.
All that’s been proven is we can apply patterns of discovery well defined by info theorists a century ago to electron state in a machine.
We’ve hardly upended immutable law established by physical experiment. Discovery of human ability to capture natures energy in its machines is also not so new; trebuchets and arrows leveraged gravity.
Only 14% of the adult public has more than a bachelors. Religion proved it is not that hard to convince the majority in titillating nonsense.
Individually none of the specific discoveries that go into SpaceX or Google were invented there. Iterating on well known ideas with well known technology is what humans do. It’s fascinating and interesting.
The figurative gibberish about a minority is to serve nation state propaganda about how amazing it is here. But inside the gulag we all defer to the minority needed to sell the illusion.
Not too surprising given the obscene cash injection they just gotten. It be a huge fumble if they didn't try to scoop up all the talent they could ever need.
Easily worth it for OpenAI. Sam is living up to everything pg always said he was.