Thread by @powerfultakes

AI RISKS & CRYPTO (or, JAIL SZN)

If you assign a high chance to AI doom (I don't, but say you do) then, come to think of it, most effective thing you could do is ban crypto. Not just regulate. Outright ban everything crypto related, on the China model (e.g. 10 years in prison).

As it stands, there's a reasonable chance that a malevolent AGI can be isolated and "unplugged". But what to do about an AGI that runs decentralized in the cloud, cryptographically secured, masked by decentralized VPN, limitless power to mount Sybil attacks on identity systems?

I explored such a scenario in my short story WAGMI in November 2021. akarlin.com/wagmi/ The AI, driven by the need to make its creators money, trades its way up, making identities and buying more processing power in a feedback loop, until it owns most of the world economy.

You're not going to go very far from just capping hardware, since normies, states, gamers, etc. all want more powerful computers, but crypto bros already have an extremely toxic reputation in most of the world & it would actually be politically feasible to repress them.

Just hook AI safety talking points into already extant rhetoric about climate impacts (Greens/left), tax evasion (left), undermining the dollar (burger boomers), "not being productive" (Munger like fossils), rogue regime sanctions evasion (neocons), extremist fundraising...

... and criminalize the possession and trade of *decentralized* cryptocurrencies.

I obviously don't advocate or endorse any of this because I'm an AI optimist, a long-term crypto bull ideologically, and would probably be personally #rekt on top. akarlin.com/ai-safety-deba…

However, from the view of intellectual honesty, I do think crypto criminalization is the lowest hanging branch aka most politically feasible way to reduce AGI risk. Appealing to private corps to forego the profit motive or nuke ASML/TSMC are... rather less realistic propositions.

It's the principle of tradeoffs and having your cake and eating it too. You have a choice, you can choose credible AI restrictionism or you can choose global cryptographic freedom & faster radical life extension timelines, I don't think you can have both.

https://twitter.com/powerfultakes/status/1639669905791254529

The Karlin Trilemma?:

* Development of AGI with all its manifold boons inc. RLE and IQ/acc
* AI safety
* Network states (financial & identity systems secured by decentralized cryptocurrencies)

Pick two!