2 Comments

https://stories.td.com/us/en/article/td-confirms-ownership-stake-in-the-charles-schwab-corporation-following-completion-of-transaction

So you can help a money launderer when you are trading securities! Yay!

"

motte-and-bailey anti-argument

"

Actually, I think Pascal's Wager fits perfectly with TESCREAL eschatology. I don't know if there's a heaven or a hell / I don't know if there's an AI post-scarcity utopia or AI revolt. But rather than risk I'm wrong... I will believe and devote myself to this idea of god / I will believe and give unfettered power and access to AI Tech Bros.

Same problem as Pascal's Wager. 1) Not free. Opportunity cost in praying / opportunity cost in listening to AI Tech Bros. 2) If wrong god chosen, real god might be pissed and lose Heaven or give you Hell / if AI Tech Bros are self serving, the real stewards to AI utopia is lost and some rogue stewards to AI might destroy the world.

If AI Tech Bros can't prove with evidence (and evidence would include both forecasting at least decently on smaller scales enough that their forecasting can/will scale up). Why make them your gods? (Assuming there could be an evidence-based way to find "god.")

Expand full comment

This sounds vague and philosophical, but there is a bit of evidence-based methodology that is meaningful. It's not sexy. Just plain ol' systems hardening.

The AI systems that do not go rogue or get exploited but follow intentions best, and have a consistent track record over a long period, are simply more trustworthy. But nothing should ever be absolutely trusted, and trust cannot be done without time, which of course dismays tech accelerationists who want a solution instantly. But hardening is not quick, or at least cannot be trusted to be unbrittle.

Expand full comment