I am highly bullish on Anthropic's tech and highly bearish on their stock valuation. It seems clear to me that models will be commodities in the near future and that most of the value will be accrued by the companies using them, not the providers.
I honestly think, given that, the best way to capture the value of AI is to buy the index fund VT
I agree. At work we use just whatever models, be it from OpenAI, Anthropic, Alphabet or even Chinese with MS Copilot, or open source agent tools. There is zero switching cost, and we just go for whatever model is cheap while providing adequate quality. Right now that's GPT Luna from OpenAI. Maybe next month it will be something from Anthropic. They are commodities.
I don't believe any government should control algorithms, although they can certainly regulate them or require more transparency. However, non-European entities could use their control over them with negative consequences for the EU and its citizens, which I think is why a disinvestment is one possible way to prohibit that.
Ideally we should have a competitive market where I can choose whether I want the china-controlled algorithm, the EU-controlled algorithm, or some unregulated algorith. Being locked-in to a single option is a big problem.
Ideally, people can choose their own content algorithm (or lack of) with no hijinks or dark patterns around defaults, selection, etc. Perhaps warnings would have to be put in place on certain algorithms, similar to a "contains excess amounts of sugar" warning or so. And some education to go with it. Second most ideal would probably be democratic oversight with reasonable rules in place, as long as the democracy is functioning. Then probably banning it, though that's a bit of a can of worms. Last place is surely people like Zuckerberg controlling it. Really anything is better than what we have now.
Ok, but this is just a statement, not a viable plan for actually running these platforms. It's clear how private tech companies are incentivized to run them, it's unproven but comprehensible how the state could be incentivized to run them.
No one has demonstrated that there's even a coherent plan to get decentralized, sovereign social media off the ground beyond just expecting millions of people to spontaneously become extremely principled. On the contrary, all the evidence throughout the history of the internet is that people will give up almost any power they have in exchange for temporary subsidies and convenience.
Never better than a bunch of robber barons? Never better than totally unregulated medicine? Never better than totally unregulated banks? Never better than letting people dump their factory's pollution directly into rivers?
reply