A fair question. I suspect OpenAI's research agenda is actually less idyllic than most think since a) they never promised to "change the world", and b) their projects eventually must serve multiple masters with vested interests.

I don't really grok OpenAI's long term mission plan. "Keeping AI open" seems rather banal to survive long amid a cultural ethos as purposeful and dynamic as SV's. Perhaps the prospect of meandering their rather random path proved less pregnant with possibilities than AK had hoped.

Or maybe Tesla promises to build more than just KITT? Musk has huge ambition. I suspect Karpathy was enraptured at the prospect of sinking his teeth deep into a wide variety of tasty pies there.

OpenAI is a weird animal. Their research seems generally less impactful than rest of the top tier labs out there. But they do have great people on the team. Is it because their engineering infrastructure is subpar?

Depends on how you see it though; the most impactful stuff is not necessarily the most reported in the media. As always, the media lags these things and so the public don't get to see shit coming.

I am pretty catching up-to-date with this field, and I didn't really get my DL(AI though it is pointless buzzword) news from media.

As far as I can remember, most important research from OpenAI is InfoGAN. Others feel somewhat below outstanding. Their open source record on the other hand is pretty solid, gym/universe and various implementations and testing of existing models does live up to the openness in their name.

I heard there were layoffs of engineers there. But obviously unrelated. OpenAI will continue unchanged.

I imagine Pieter Abbeel and the rest of the gang will be able to carry on in his absence.

This. Abbeel is a highly underrated research powerhouse.

I'm curious why you think Abbeel is underrated. He is one of the few ML researchers to not succumb to the pull of the industry (at least completely) and his work is highly respected in the community, especially continuous control.

Doesn't OpenAI gets large part of their funding from Elon?

I mean OpenAI is not a competitor to Tesla.

But it very well might be a beneficiary of Tesla's Autopilot machine learning work. Mr Musk is not shy about sharing tech within his companies (Tesla adopted friction stir welding techniques pioneered at SpaceX [1]). Also, the patents Tesla offered to license to other automakers [2] to help accelerate the shift to electrified transportation.

Now, granted, SpaceX does not file patents because their competition is primarily nation-states [3], but I feel thats a valid concern with regards to that domain.

"We have essentially no patents in SpaceX. Our primary long-term competition is in China," said Musk in the interview. "If we published patents, it would be farcical, because the Chinese would just use them as a recipe book."

[1] https://electrek.co/2015/05/24/spacex-transferred-novel-weld...

[2] https://www.tesla.com/blog/all-our-patent-are-belong-you

[3] http://www.businessinsider.com/elon-musk-patents-2012-11

Friction stir is really old... how did they pioneer it? Perhaps they perfected it?

Not friction stiring itself, a specific technique.

OpenAI is pretty independent and has grown quite considerably. Karpathy was a founding research scientist but it seems like the nonprofit will march on just fine