Jon Lok

I like funny things

[Jensen Huang @JensenHuang]

For my first post, I’m sharing a letter @NVIDIA signed on why open models matter.

AI will transform every industry, power every company, and be built by every country.

Open models strengthen safety and cybersecurity, accelerate innovation and diffusion, and enable sovereignty.

The world needs both frontier closed models and frontier open models.

https://t.co/AUKzoQ5Ikb

[Jon Lok @JonLok18]

@kimmonismus I think that they'd have to go through a due process though and prove that this actually happened. Would they not?

[Jon Lok @JonLok18]

@tylerangert That looks pretty

[Jon Lok @JonLok18]

@StirlingWisdom lol no. you try and go into a public bathroom with a bag much easier and sanitary not to

[Jon Lok @JonLok18]

@PeterDiamandis …

[Jon Lok @JonLok18]

@elonmusk Reminds me of the trans issue, first they denied surgery was being performed on minors then when the evidence came out, they said it was a good thing!

[Jon Lok @JonLok18]

@tunguz I would assume all the valuable stuff sits in ITAR and therefore not relevant

[Jon Lok @JonLok18]

@teortaxesTex @paulmarin90 Not sure, he does open source quite a bit both Tesla and spacex

[Jon Lok @JonLok18]

@teortaxesTex Why can’t he do that in your view?

[Jon Lok @JonLok18]

@martianwyrdlord @curtis_yarvin Make it happen and more !

[Jon Lok @JonLok18]

I'm not sure Elon really cares. His job is to put as much pressure as possible on his team and they keep making him look bad, which allows him to keep more pressure on to deliver something that's extremely difficult. Whether he contains the message the way you want him to do it, I think, has no bearing on the fundamentals of the business. Either they're going to deliver on time or they are not.

[Jon Lok @JonLok18]

@alojoh agreed

[Jon Lok @JonLok18]

@ritwikpavan I just want it to count reps nothing more

[Jon Lok @JonLok18]

@rdominguezibar What matters most is having a valuable product you can copy

[Jon Lok @JonLok18]

@teortaxesTex Treasonous courage

[Jon Lok @JonLok18]

The trillions in AI capex are inference-driven. Training the next big model is in the 10s of billions?. We could fund that without closed-model profits, then let others run the services. Might take longer but we would get there irrespective.

The big labs have monopoly ambitions (god complex) and that’s why you see them rushing to capture everything. Not sure it’s a good idea for the intelligence maker to control so much, I think it would be healthier for them to have calmer ambitions.

[Jon Lok @JonLok18]

@Suhail They are distilling, that’s mainly what they have been doing for the last 30 years. Ripping off western tech and making it more efficient and cheaper.

[Jon Lok @JonLok18]

@iamgingertrash Mate we are entering the age of abundance and your crying about style…. It is inevitable and the US will get credit for it either way. Cultural victory was won in 1776

[Jon Lok @JonLok18]

@bigeagle_xd keep going!

[Jon Lok @JonLok18]

Still, in the end, open-weight models deter further AI capex.

Yes and no, the evidence so far shows no but conceptually when the open models are good enough it certainly could slow capex spend down.

However, it is worth to seperate out what part of capex spend actually decels. Training capex relative to inference is widening by the day. I do not belive that even if you get par open models that labs will slow training capex because it’s not actually that much in the grand scheme of things.