Discussion
It's been obvious for a few years that the AI industry is building a great big
financial bomb made of recycled debt, but I'm sort of amazed at just *how* big the bomb is.
Even if AI was *all that* technically - which it isn't - the financial bomb will *still* go off.
At this point the questions should all be: when, not if, it goes off - what happens next?
@davidgerard From a comforting first class steward in 1912: no need to worry, ma'am, you can go back to bed. Just a little iceberg.
@davidgerard I have that same question, plus some more.
I think that probably it won't blast as a bomb, but that will act a long term toxic chemical, tainting everything and everyone.
My first question is about financial institutions funding it. How comes they're still putting money in a debt that will maybe just repay interests, but never the capital?
This is blatant nonsense even for the man in the street.
@davidgerard - and, extremely crucially, who will be held responsible & stripped of their riches? Because if They™️ are allowed to remain free & wealthy, then the whole operation was a success from their perspective & they'll have no reason not to immediately embark on the next journey of fraud & destruction...
Importantly, rich people typically get richer when there’s a crash. A crash brings down the value of a lot of things and lets anyone with liquidity buy up assets.
Now, the really amusing thing would be if the EU took an idea from the Chinese playbook and, when the crash happens, minted a few billion Euros specifically to buy undervalued US assets for sovereign wealth funds. This wouldn’t cause inflation because the money supply would reflect the value of the assets.
@david_chisnall @davidgerard - and/or from the Icelandic playbook of making the perpetrators pay to protect the population, rather than the reverse...
That would be hilarious.
@davidgerard Whatever it is,DAO 216-26 seems like an indicator that the Department of Commerce doesn't have a carpet big enough to hide the consequences of it bursting.
the most important humans have their well equipped bunkers - the others can look into their future as soylent green
@davidgerard Just another doomsday prophecy
@davidgerard
I expect a paralell to the 2008 crisis:
There's so much wealth tied in these companies, many will be bailed out by the state(s). Massive redistribution from the bottom to the top.
To the sheer financial to big to fail there's the ai companies sold themselves as important industry, another motive for the state to step in.
@niedlichenacktschnecke the money doesn't exist to "bail out" OpenAI and Anthropic. I mean they could declare they're printing $1t, or they could not do that. Who will structurally *need* bailing out?
or maybe i'm assuming governmental competence not in evidence
@davidgerard @niedlichenacktschnecke What happens next depends on who is in power in the US. If Rs they will try to bail out the unbaleable by printing money. If the D socialists get power these companies with be fucked.
@davidgerard I am not sure a lot happens beyond a lot of imaginary money for imaginary data centres full of imaginary chips vanishes The people who put the money in at the bottom cash out fine, the people who were late and index funds don't.
If you loan someone 300 million to buy chips as yet unmade and they give you the 300 million back in orders and then go bust providing you got the collateral against the loan right nothing actually moves.
@davidgerard It reminds me to check my superannuation to see if my fund has any AI slop companies in the mix.
@davidgerard my expectation is that it ends in the usual way: people loose money, then they move on, and use the technology like for things they don't understand. The impact of this technology will be a lasting one, and few people will be aware of epistemological challenges, as it always is when a technology gets normal.
Install bonfire.cafe
Get the full app experience
@davidgerard I'd be curious if it wipes out the entire ML sphere by making investors bail immediately when they get a whiff of "AI" in the future, and that's primarily the fault of ML people that either supported or ignored it when the woo-woo salespeople decided to wrap everything up into one big blob of "AI".
They're doing "yeah it's uh...it's like Claude, you like Claude, right?" for anything now, and I'm not sure if the flip to "No, it's NOT like that, it's way more mundane" will work