• Moderators, please send me a PM if you are unable to access mod permissions. Thanks, Habsy.

OT: World Politics


Japan is about to quietly go nuclear and pivot to improving relations with China. A bunch of Asia experts have suggested that this was their back up plan if Trump got re elected and tore down the western order. The Japanese are sober pragmatists at heart and if US assistance can't be relied upon, knew that better relations with China but while having nuclear deterrence was the only viable path forward.

Fuck this timeline.
 
We have a billion years worth of thorium in Canada too. Shit doesn’t work so well right now at power generation.

There's some encouraging work being done on the corrosion and inspection issues and I could see this being a thing that exists in the world in 30 years. But we also might have fusion by then, so the fuck would we need safer fission for?
 
Perun (who is a fucking legend and anyone into longer form youtube content on the Ukraine war should give his stuff a watch) dropped his breakdown of what a NATO without the US looks like. Capabilities, challenges, etc

TLDW: Better than you think once they solve some logistics challenges of tying multiple smaller forces together and actually pretty fucking competent if Ukraine makes it out of this war as a functioning European nation. They already spend a combined bundle on defence. Their combined forces and spend dwarf Russia.

I've started it where the bulk of the most pertinent analysis starts, around the 23 minute mark.


View: https://youtu.be/7giYIisLuaA?si=QeKhynnAVZBMZzB9&t=1381
 
Every time I see someone question Europe's tech competitiveness because the US is so much further ahead in "AI", they're telling on themselves as deeply unserious.

LLM AI is a 400 Billion dollar capex spend in frantic search of a multi trillion dollar problem it solves that would justify it's existence. Mewling about missing out on blowing 100's of billions developing AI models is going to end up a lot like complaining in 1990 about not spending hundreds of millions on developing competing internet protocols. In the end, there was no money in that, the winning protocol became a near worthless commodity and the winners at the time were the PC hardware and software manufacturers who bridged the gap from the tech to the consumer.

The AI "industry" barely even exists. The most successful company in it burns 5 Billion a year developing a product that the best monetization to date is a $200/mo subscription service that....loses....money per subscriber because the computing cost required to deliver these results is crazy high. This is like Tesla with self drive all over again where the valuation of the company is only justified by a promise that they'll invent a tech that doesn't exist, & might be impossible to invent. That is the AI industry with AGI right now. Professional tech nutcases like Sam Altman have promised that AGI is right around the corner if we just throw a few trillion dollars at the problem. In the interim, we're just going to jam our shitty, expensive LLM's into everywhere we can to justify the ongoing expense necessary to get to AGI and just like Elon with self drive, they're waiting to pass some stupid fucking self imposed threshold that isn't actually AGI before doing a touchdown dance and proclaiming that they've done it and then force it on the world as if it's the singularity.

There are definitely some commercial use cases for LLM AI, and it can create billions of dollars of growth, efficiency, etc for numerous industries but the money is all going to be in the Apps, the SAAS businesses that use it well, etc. Small peanuts compared to the epic spend, and even that is dependent upon the models ever becoming energy efficient enough to be economical.
 
Every time I see someone question Europe's tech competitiveness because the US is so much further ahead in "AI", they're telling on themselves as deeply unserious.

LLM AI is a 400 Billion dollar capex spend in frantic search of a multi trillion dollar problem it solves that would justify it's existence. Mewling about missing out on blowing 100's of billions developing AI models is going to end up a lot like complaining in 1990 about not spending hundreds of millions on developing competing internet protocols. In the end, there was no money in that, the winning protocol became a near worthless commodity and the winners at the time were the PC hardware and software manufacturers who bridged the gap from the tech to the consumer.

The AI "industry" barely even exists. The most successful company in it burns 5 Billion a year developing a product that the best monetization to date is a $200/mo subscription service that....loses....money per subscriber because the computing cost required to deliver these results is crazy high. This is like Tesla with self drive all over again where the valuation of the company is only justified by a promise that they'll invent a tech that doesn't exist, & might be impossible to invent. That is the AI industry with AGI right now. Professional tech nutcases like Sam Altman have promised that AGI is right around the corner if we just throw a few trillion dollars at the problem. In the interim, we're just going to jam our shitty, expensive LLM's into everywhere we can to justify the ongoing expense necessary to get to AGI and just like Elon with self drive, they're waiting to pass some stupid fucking self imposed threshold that isn't actually AGI before doing a touchdown dance and proclaiming that they've done it and then force it on the world as if it's the singularity.

There are definitely some commercial use cases for LLM AI, and it can create billions of dollars of growth, efficiency, etc for numerous industries but the money is all going to be in the Apps, the SAAS businesses that use it well, etc. Small peanuts compared to the epic spend, and even that is dependent upon the models ever becoming energy efficient enough to be economical.
Alrighty, then. Fuck load to absorb on the fly, but that's okay. Notes & time appreciated, as always. 👍
 
Back
Top